                                                                  Computer Standards & Interfaces 97 (2026) 104118


                                                                      Contents lists available at ScienceDirect


                                                            Computer Standards & Interfaces
                                                               journal homepage: www.elsevier.com/locate/csi




Post-quantum PAKE over lattices revised: Smaug-T.PAKE for mobile devices
Kübra Seyhan a                 ,∗, Sedat Akleylek b,c              , Ahmet Faruk Dursun a
a
    Department of Computer Engineering, Ondokuz Mayis University, Faculty of Engineering, Samsun, 55139, Turkiye
b
    Chair of Security and Theoretical Computer Science, University of Tartu, Institute of Computer Science, Tartu, 50090, Estonia
c Department of Software Engineering, Istinye University, Faculty of Engineering and Natural Sciences, Istanbul, 34396, Turkiye




ARTICLE                  INFO                                ABSTRACT

Keywords:                                                    In this paper, an efficient post-quantum secure password-authenticated key exchange (PAKE) scheme from
Post-quantum cryptography                                    a well-structured lattice-based key encapsulation mechanism (KEM) is proposed. The generic KEM to PAKE
Lattice-based cryptography                                   idea, OCAKE, is modified by considering hybrid module learning with errors (MLWE) + module learning
SMAUG-T
                                                             with rounding (MLWR) assumptions to obtain explicit password-based authentication from SMAUG-T.KEM
Password-authenticated key exchange
                                                             procedures. As a KEM primitive, SMAUG-T.KEM is chosen due to its performance against the National Institute
                                                             of Standards and Technology (NIST) standard Crystals-Kyber (Kyber) to obtain an efficient and post-quantum
                                                             secure PAKE scheme. Firstly, the anonymity and fuzziness properties of SMAUG-T.KEM are proven to fit
                                                             the OCAKE approach in constructing the PAKE version of Smaug.KEM. Then, the post-quantum security of
                                                             the proposed SMAUG-T.PAKE is analyzed in the universal composability (UC) model based on the hybrid
                                                             security assumptions and proved properties. The reference C and JAVA codes are written to evaluate whether
                                                             the targeted efficiency is achieved in different platforms. Based on the central processing unit (CPU) and
                                                             memory usage, run time, and energy consumption metrics, the proposed solution is compared with current
                                                             PAKE proposals. The performance results showed that SMAUG-T.PAKE, with two optional encryption modes,
                                                             Advanced Encryption Standard (AES) or Ascon, presents better performance than the other module-based PAKE
                                                             solutions from lattices in terms of both reference and mobile results.



1. Introduction                                                                                        The computational hardness of traditional PAKEs was basically
                                                                                                   captured by following the hardness of discrete logarithm problem
    A PAKE protocol provides secure key sharing on an insecure channel                             (DLP). In the pre-quantum era, the security of PAKEs is maintained
by using a pre-shared password as an authentication component [1].                                 as there is no efficient and polynomial-time algorithm to solve DLP
In recent years, PAKE protocols have been preferred in wireless com-                               if the suitable parameter set is selected. The first appearance of the
munication, e-passports, and the Internet of Things, where efficiency,                             Shor algorithm [7] changed this situation and started a new challenge
portability, simplicity, and independence are essential [2,3]. PAKE                                for public key cryptography (PKC) in the age of large-scale quantum
usage in resource-limited communication models establishes indepen-                                computers. NIST announced a call to be ready for the post-quantum
dent, secure, and portable authenticated communication without extra                               era by determining PKC standard(s) in 2016. In this period, it was
public key infrastructure, complex components, or central authority.                               aimed to select the quantum-resistant digital signature and public-key
The main trade-off in the PAKE protocols arises between efficiency                                 encryption&key-establishment algorithms. The NIST standardization
and security since they use low-entropy pre-shared passwords to de-                                process was finalized in 2024 and 2025, and lattice-based CRYSTALS-
rive high-entropy shared keys. The first introduction of PAKEs to the                              KYBER and code-based HQC were determined as the post-quantum
literature was given with encrypted key exchange (EKE) by Bellovin                                 KEM standard, while lattice-based CRYSTALS-Dilithium and Falcon,
and Merritt in 1992 [4]. After this first attempt, several PAKE solutions                          and hash-based SPHINCS+ were determined as digital signature stan-
were proposed, and standardization attempts were started by covering                               dards [8]. In addition to the international call of the NIST, China,
different design settings, primitives, properties, and usage areas. The                            Korea, Ukraine, and Russia also started initiatives to determine their
recent standardization effort was made by the Internet Engineering                                 standards for post-quantum cryptography (PQC). Even if there has
Task Force (IETF) in 2020, and Cpace and OPAQUE were announced                                     not been started a standardization process specific to PAKEs, there
as standard traditional PAKE protocols [5,6].                                                      is ongoing research on the post-quantum secure PAKE design process


     ∗ Corresponding author.
       E-mail address: kubra.seyhan@bil.omu.edu.tr (K. Seyhan).

https://doi.org/10.1016/j.csi.2025.104118
Received 25 August 2025; Received in revised form 29 October 2025; Accepted 12 December 2025
Available online 15 December 2025
0920-5489/© 2025 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
K. Seyhan et al.                                                                                                Computer Standards & Interfaces 97 (2026) 104118


in the literature and industry. In particular, PQC algorithms have                  • In order to evaluate the effect of KEM and encryption/decryp-
higher key and cipher sizes compared to traditional cryptosystems,                    tion operations on PAKE security analysis, MLWE+MLWR-based
and they require high bandwidth and system resource usage, making                     adaptations are also followed in the UC model.
efficient PQC protocol design difficult. As it is known, after standard             • Detailed security analysis of the proposed PAKE based on the
cryptographic principles are determined, the needs are met by adapting                modified security definitions shows that SMAUG-T.PAKE main-
them to different purposes and application areas as a black box in                    tains post-quantum security even with the additional components
design. So, proposing a post-quantum secure PAKE based on a well-                     it includes.
studied algorithm will be one of the appropriate candidates for the                 • The efficiency analysis of the proposed PAKE is done by con-
future security of PKCs. It is necessary to provide the required security             sidering different application areas and additional primitive us-
definitions for the combination of different cryptographic principles in              ages. Additionally, detailed performance analyses are provided
PAKE design, to present evidence on ensuring post-quantum security,                   to demonstrate that more efficient structures can be constructed
and to perform performance evaluations on different platforms. Ad-                    compared to PAK-based PAKE construction models.
ditionally, comparative performance and efficiency analysis with the                • The constructed SMAUG-T.PAKE is implemented in C for gen-
methods used for PAKE design in the literature and the effect of using                eral use-case performance and also in JAVA for mobile appli-
different models in cryptographic primitives used as black boxes should               cation usage. To make a meaningful comparison, NIST standard
be examined to evaluate post-quantum secure PAKE protocols.                           CRYSTALS-Kyber is also implemented in C, based on the same
                                                                                      construction idea.
1.1. Motivation                                                                     • The efficiency of SMAUG-T.PAKE is analyzed in terms of CPU
                                                                                      usage, memory usage, and energy consumption.
    PQC has received tremendous interest in industry and academic                   • According to the comparison, the constructed PAKE provides bet-
research due to the increasing work to build a sufficiently powerful                  ter performance results than other module-based PAKE solutions.
quantum computer. To be ready for the post-quantum era, active work                   Moreover, the proposed SMAUG-T.PAKE gives the best results
is being carried out with various attempts, such as standardization                   when using the lightweight cipher Ascon.
efforts, literature research, and industrial initiatives, specifically re-
garding the essential key agreement requirement. Among key sharing               1.3. Related work
schemes, PAKEs stand out with their simple operations and strong
security based on passwords. While studies on standardization and im-                In the literature, different PAKE design methodologies have been
provement of traditional PAKE protocols continue, the need to evaluate           proposed to be ready for the post-quantum era. The presented lattice-
the post-quantum effect has also emerged. Although standardization               based solutions were generally constructed by combining different
efforts have been initiated to determine post-quantum secure standards           projective hash functions (PHF), reconciliation structures, password-
for basic public key cryptosystems, no attempt has yet been made for             related computations, etc., according to the selected design idea. The
specialized primitives like PAKEs. Although there has been increasing            proposed solutions can be divided into three categories: Use projective
interest in developing quantum-safe PAKEs in recent years, little work           hash functions to convert an encryption scheme to a PAKE [14–20],
has yet been done on building PAKEs by integrating standard PQC                  build a PAKE from a key exchange (KE) idea by adding password-based
algorithms. The main motivation for this paper comes from [2,9,10],              authentication [21–29], and convert a KEM to PAKE [12,13,30–34]. In
which identified the design, security, and performance analysis of a             Table 1, a snapshot of lattice-based PAKE literature is given.
post-quantum secure, relatively efficient PAKE derived from standard                 It is known that KEM schemes contain some extra encryption/
primitives as an open problem. The cryptographic primitives in con-              decryption procedures, functionalities, and computations to ensure
structing the efficient post-quantum secure PAKE solution must be                strong security properties. So, the nature of KEM-based PAKE schemes
evaluated for their effects on security analysis and performance.                tends to show poor performance results and strong security features.
                                                                                 In the lattice-based PAKE literature, two basic models for converting
1.2. Contribution                                                                a KEM to PAKE exist. The first model combines traditional PAKE de-
                                                                                 sign [1] and well-structured KEM to add password-based authentication
    In this paper, we focus on the problem of whether it is possi-               to the KEM idea. It uses KEM procedures as a black box and adds
ble to construct efficient and post-quantum PAKE using well-defined              password-related primitives to provide password-based authentication
post-quantum KEM algorithms and additional primitives. To obtain an              in the KEM. In the second model, four generic KEM to PAKE construc-
efficient PAKE, the SMAUG-T [11] algorithm, which is the Korean PQC              tions were provided in the sight of traditional PAKE designs. These
KEM standard and has been shown to be more efficient than the NIST               generic PAKE models contain extra ideal cipher operations and use
KEM standards, was used to improve efficiency. With the focus on                 KEM procedures to ensure explicit or implicit authentication.
efficiency, the idea of OCAKE [12] is adopted into the proposed model                As the main focus of this paper, one-round lattice-based PAKE pro-
to capture explicit password-based authentication. The proposed PAKE,            posals that were constructed using KE and KEM approaches by adding
SMAUG-T.PAKE, is the first MLWR+MLWE-based PAKE scheme that                      password-related components will be summarized as follows. Note that
provides efficiency and post-quantum security. The contributions of this         to analyze the performance of the proposed PAKE, MLWE/MLWR-
paper to the literature can be summarized as follows.                            based PAKE construction will be used. In the following part, the main
                                                                                 construction idea of these PAKEs will be briefly summarized.
    • The first MLWE+MLWR-based PAKE protocol that provides ex-                      In [35], the first lattice-based construction of conventional
      plicit authentication, anonymity, fuzziness, and efficiency for            password-authenticated key exchange (PAK) PAKE [1] was presented
      post-quantum era security is proposed.                                     in the literature regarding ring learning with errors (RLWE) hardness
    • As an extended and modified version of [13], more efficient                assumptions. The proposed PAKE provided a password-authenticated
      lattice-based PAKE with reduced encryption/decryption and                  shared key generation for the two-party communication model. The
      Smaug.KEM operations are constructed. Unlike [13], MLWE+                   security of this scheme was analyzed using the ROM assumptions.
      MLWR-based security definitions are defined in the UC model to             In [23], an efficient MLWE-based PAKE scheme, MLWE.PAK.PAKE,
      capture explicit authentication.                                           was proposed by considering the traditional PAK design idea [1].
    • Firstly, the anonymity and fuzziness properties analysis of                MLWE.PAK.PAKE scheme consists of one round and was designed for
      SMAUG-T.KEM are defined and provided to build modified                     the two-party communication model. The security analysis against
      MLWR+MLWE-based PAKE construction.                                         password dictionary attacks was performed on the ROM. In [25],

                                                                             2
K. Seyhan et al.                                                                                                      Computer Standards & Interfaces 97 (2026) 104118


                      Table 1
                      Outlook on lattice-based PAKEs.
                        PAKE construction       Literature          Hardness               Security            Additional
                        model                   proposals                                  model               primitives
                                                [21]                RLWE                   ROM                 X
                                                [22]                RLWE                   ROM                 X
                                                [23]                MLWE                   ROM                 X
                                                [24]                RLWE                   ROR                 X
                        PAKE from KE
                                                [25]                MLWR                   ROM                 X
                                                [26]                RLWE                   ROM                 X
                                                [27]                RLWE                   ROR                 X
                                                [28]                RLWE                   ROM                 X
                                                [29]                MLWE                   ROM                 X
                                                [30]                MLWE                   ROM                 X
                                                [31]                MLWE                   UC                  Feistel
                                                                                                               construction
                                                [32]                MLWE                   ROM                 Oneway
                        PAKE from KEM
                                                                                                               plaintext-checking
                                                [12]                MLWE                   UC                  Ideal cipher
                                                [13]                MLWE                   UC                  Two Cipher
                                                                    + MLWR
                                                [33]                Non-uniform LWE        ROM                 Ideal Cipher
                                                                                                               XOR
                                                [34]                X                      ROM                 SPHF
                                                                                                               XOR
                                                                                                               Encryption
                                                Ours                MLWE                   UC                  Cipher
                                                                    + MLWR
                                                [14]                LWE                    ROM                 SPHF
                                                                                                               encryption
                        PAKE with PHF           [15]                RLWE                   UC                  Oblivious Pseudorandom
                                                                                                               function
                                                                                                               NIZK
                                                [17]                LWE                    Standard model      ASPHF
                                                                                                               Encryption
                                                                                                               ECC
                                                [18]                LWE                    ROM                 ASPHF
                                                [20]                LWE+LWR                ROM                 SPHF
                                                                                                               Password hashing scheme
                                                [19]                LWE                    ROM                 SPHF
                                                                                                               NIZK

                      –ROM: Random Oracle Model –ECC: Error Correction Code –ROR: Real or Random –NIZK: Non-Interactive Zero Knowledge
                      –UC: Universal Composability –SPHF: Smooth PHF –ASPHF: Approximate SPHF.


MLWR-based PAKE was constructed by considering NIST’s Saber [36]                   idea was analyzed using the ROM assumptions. In [33], KEM’s PAKE
KE idea and traditional PAKE approach [1]. Even if the design structure            design utilizes public-key authentication using XOR instead of public-
of the proposed Saber.PAK.PAKE is the same as MLWE.PAK.PAKE, it                    key symmetric encryption. The goal is to create perfect privacy schemes
presented better performance results due to the efficient structure of the         by eliminating the need for ideal passwords. In [34], the definitions
MLWR problem. In [30], the PAKE version of the NIST KEM standard,                  for PAKE models based on post-quantum KEM and traditional PAKE
Kyber, was proposed using KEM functionalities and basic password                   assumptions were provided. Details on the security analysis of the
components. The core construction of Kyber.PAK.PAKE is based on                    proposed hybrid PAKE models in the UC-model were also presented.
the traditional PAK approach to capture password-authenticated shared              In [12], Beguinet et al. proposed two different PAKE construction
key generation. In [29], a MLWE-based three-party PAKE protocol was                models that include KEM as a black box. According to the provided
proposed by following the KE to PAKE design idea. In the security                  generic models, the CAKE version presents password-based implicit
analysis, ROM-based assumptions were followed to provide the formal                authentication with strong security proofs based on adaptive corrup-
security.                                                                          tions even if the receiver is unsure that he/she can receive the session
    In recent years, researchers have been working on how to convert               key. The OCAKE approach captures explicit authentication using a
a well-structured KEM into a PAKE. The main reason behind this idea                key-confirmation tag without extra encryption. These two construc-
is to create PAKE versions of standard KEMs for different applications             tions require a KEM that provides fuzziness and anonymity properties
or usage areas. Different models, [12,31–34], have been proposed in                for the public key and ciphertext, respectively. The security analysis
the literature considering the lattice assumptions. In [31], a compact             was provided based on Kyber’s assumption by considering password
PAKE construction, which requires the underlying KEM’s one-wayness                 authenticated-based ideal functionality in the UC model.
and anonymity, and the public key’s uniformity, was defined. As a
case example, Kyber-based PAKE was proposed and analyzed in the                    1.4. Outline
UC model. In [32], a generic PAKE model from a one-way secure
against checkable attack secure KEM was introduced. The proposed                       In Section 2, the mathematical background is summarized. In Sec-
construction also required that underlying KEM’s anonymity and fuzzi-              tion 3, the constructed SMAUG-T.PAKE is defined, and the correctness
ness. By considering Kyber KEM functions, the security of the proposed             analysis is given. In Section 4, the proof of security is detailed under the

                                                                               3
K. Seyhan et al.                                                                                                                 Computer Standards & Interfaces 97 (2026) 104118


                   Table 2
                   Notation.
                    𝑞, 𝑝                    :        Positive integers modulo, power of 2.
                    Z𝑞                      :        The quotient ring of integers in modulo 𝑞.
                    Z𝑞 [𝑥]                  :        Polynomials with coefficients in Z𝑞 .
                    ℜ𝑞 = Z𝑞 [𝑥]∕(𝑥𝑛 + 1)    :        Quotient ring.
                    ℜ𝑘×𝑘
                       𝑞
                                            :        Ring of 𝑘 × 𝑘 matrices on ℜ𝑞 .
                    𝜒                       :        Discrete Gaussian distribution with 𝜎 = 1.0625 standard deviation.
                    𝑆𝜂                      :        SMAUG-T.KEM secret key and error term distribution.
                                                     For 𝜂 ∈ Z, 𝑆𝜂 denotes the set of polynomials of degree less than n with coefficients in [−𝜂, 𝜂] ∩ Z.
                    𝑇                       :        Transpose.
                    XOF                     :        Extendable function. Utilized to generate the seed of A and set as Shake-128.
                    expandA                 :        Sampling function for SMAUG-T.KEM’s general public key.
                                                     expandA(⋅): 𝑝 ∈ {0, 1}256 → 𝐴 ∈ ℜ𝑘×𝑘   𝑞
                                                                                               .
                    ℎ𝑤𝑡ℎ                    :        Sampling function of SMAUG-T.KEM’s sparse polynomials with hamming weight ℎ.
                                                     ℎ𝑤𝑡ℎ (⋅) → 𝑆𝜂𝑘 .
                    ℎ𝑠, ℎ𝑟                  :        Non-zero coefficients of sparse polynomials, where
                                                     ℎ𝑠 = {140, 150, 145} and ℎ𝑟 = {132, 147, 140} for 128, 192, 256-bit security.
                    𝑡                       :        Constant rounding component, where 𝑡 = 2.
                    𝐻                       :        {0, 1}256 → {0, 1}256 . Set as SHA3-256.
                    𝐺, 𝑘𝑑𝑓                  :        {0, 1}256 × {0, 1}256 → {0, 1}256 . Set as SHA3-512 and Shake-256, respectively.
                    ⌊⋅⌉                     :        Rounding operator.
                    𝑥 ←𝑟 𝑋                  :        𝑥 is selected uniformly random from 𝑋 distribution.
                    sk-pk-ssk-ct-pw         :        Secret key-public key-shared secret key-ciphertext-password.
                    𝛿                       :        Decryption failure probability of SMAUG-T.KEM.
                    || ⋅ ||∞                :        Infinity norm.
                    negl(⋅)                 :        Negligible function
                    𝐻1 , 𝐻2                 :        {0, 1}∗ → {0, 1}256 . Set as SHA3-256.




Table 3
Algorithms for SMAUG-T.KEM.

             SMAUG-T.PKE.KeyGen                                       SMAUG-T.PKE.Enc                                                        SMAUG-T.PKE.Dec
     seed← {0, 1}256                       Input: pk: (𝑝, 𝑏)                                               Input:   sk: 𝑠
     (𝑝, 𝜏) ←XOF(seed)                     Input: Message 𝜇 ∈ {0, 1}256                                    Input:                       𝑘
                                                                                                                    ct: (𝑐1 , 𝑐2 ) ∈ ℜ𝑝 × ℜ𝑝
     𝐴 ←expandA(𝑝)∈ ℜ𝑘×𝑘     𝑞             Input: Seed 𝑝′ ∈ {0, 1}256                                               𝜇 ′ = ⌊ 𝑝𝑡 (𝑐2 + 𝑐1𝑇 𝑠)⌉ ∈ ℜ𝑝
 sk: 𝑠 ←ℎ𝑤𝑡ℎ𝑠 (𝜏)∈ 𝑆𝜂𝑘                            𝑟 ←ℎ𝑤𝑡ℎ𝑟 (𝑝′ )∈ 𝑆𝜂𝑘                                      Message: Return (𝜇 ′ )
     𝑒 ←𝜒(𝜏)∈ ℜ𝑘                                  𝑐1 = ⌊ 𝑞𝑝 𝐴𝑟⌉ ∈ ℜ𝑘𝑝
     𝑏 = −𝐴𝑇 𝑠 + 𝑒 mod 𝑞                          𝑐2 = ⌊ 𝑞𝑝 (𝑏𝑇 𝑟 + 𝑞𝑡 𝜇)⌉ ∈ ℜ𝑝
 pk: (𝑝, 𝑏) ∈ {0, 1}256 × ℜ𝑘𝑞              ct:    Return ((𝑐1 , 𝑐2 ))
     Return ((𝑝, 𝑏), 𝑠)
           SMAUG-T.KEM.KeyGen()                              SMAUG-T.KEM.Encap(𝑝𝑘 = (𝑝, 𝑏))                          SMAUG-T.KEM.Decap(𝑝𝑘 = (𝑝, 𝑏), 𝑠𝑘 = (𝑠′ , 𝑑), 𝑐𝑡 = (𝑐1 , 𝑐2 ))
     ((𝑝, 𝑏), 𝑠′ )= SMAUG-T.PKE.KeyGen()          𝜇 ← {0, 1}256                                                      𝑚′ =SMAUG-T.PKE.Dec(𝑠′ , (𝑐1 , 𝑐2 ))
     d← {0, 1}256                                 (𝑐1 , 𝑐2 )=SMAUG-T.PKE.Enc((𝑝, 𝑏), 𝜇; 𝐺(𝜇, 𝐻((𝑝, 𝑏))))             (𝑐1′ , 𝑐2′ )=SMAUG-T.PKE.Enc((𝑝, 𝑏), 𝜇 ′ ; 𝐺(𝜇 ′ , 𝐻((𝑝, 𝑏))))
 sk: 𝑠 = (𝑠′ , 𝑑) ∈ 𝑆𝜂𝑘 × {0, 1}256               𝐾 = 𝑘𝑑𝑓 (𝜇, 𝐻((𝑐1 , 𝑐2 ))) ∈ {0, 1}256                             if (𝑐1 , 𝑐2 )≠(𝑐1′ , 𝑐2′ )
     Return ((𝑝, 𝑏), 𝑠)                    ssk:   Return ((𝑐1 , 𝑐2 ), 𝐾)                                   ssk:            Return 𝐾 ′ = 𝑘𝑑𝑓 (𝑑, 𝐻((𝑐1 , 𝑐2 )) ∈ {0, 1}256
                                                                                                                     else
                                                                                                           ssk:            Return 𝐾 ′ = 𝑘𝑑𝑓 (𝜇, 𝐻((𝑐1 , 𝑐2 )) ∈ {0, 1}256




UC framework. In Section 5, the implementation results and detailed                           As summarized in Table 3, SMAUG-T KeyGen processes follow
discussion are presented. Finally, in Section 6, the conclusion and                         MLWE assumptions to generate public and secret key pairs.
future works are explained.
                                                                                            Definition 1 (MLWE Problem [37]). Let {𝑞, 𝜂, 𝑘} ∈ Z+ , general public
2. Preliminaries                                                                            key 𝐴 ←𝑟 ℜ𝑘×𝑘                        𝑘                      𝑘
                                                                                                          𝑞 , secret key 𝑠 ←𝑟 𝑆𝜂 , and error term 𝑒 ←𝑟 𝑆𝜂 . MLWE
                                                                                                                                              𝑘×𝑘   𝑘×1
                                                                                            distribution is generated with (𝐴, 𝑏 = 𝐴𝑠 + 𝑒) ∈ ℜ𝑞 × ℜ𝑞 .
   In this section, we recall the underlying primitives basics and
security-related details. The specific symbols and abbreviations are                           The hardness of MLWE is defined by the advantage (Adv) of adver-
given in Table 2.                                                                           sary (𝐀) to solve decisional-MLWE;
                                                                                                                 |
2.1. SMAUG-T primitives                                                                        AdvMLWE                           𝑘×𝑘        𝑘
                                                                                                  𝑛,𝑞,𝑘,𝜂 (𝐀) = ||Pr[𝐛 = 1|𝐴 ←𝑟 ℜ𝑞 ; 𝑏 ←𝑟 ℜ𝑞 ;

                                                                                                                                  𝐛 ← 𝐀(𝐴, 𝑏)]−
    In the proposed PAKE, generic OCAKE KEM to PAKE construc-
                                                                                            Pr[𝐛 = 1|𝐴 ←𝑟 ℜ𝑘×𝑘       𝑘        𝑘
                                                                                                            𝑞 ; 𝑠 ← 𝑆𝜂 ; 𝑒 ← 𝑆𝜂 ; 𝑏 ← 𝐴𝑠 + 𝑒;
tion [12] is followed to obtain an efficient PAKE version of SMAUG-
                                                                                                                                   |
T.KEM [11], a Korean PQC standardization algorithm. Based on MLWR                                           𝐛 ← 𝐀(𝐴, 𝑏 = 𝐴𝑠 + 𝑒)]| < 𝑛𝑒𝑔𝑙(𝑛)
                                                                                                                                   |
and     MLWE      assumptions,    SMAUG-T.KEM        scheme     satisfies
                                                                                            In SMAUG-T’s Enc and Dec procedures, given in Table 3 the ciphertexts
indistinguishability under chosen plaintext attacks (IND-CPA) and in-
                                                                                            are generated according to the MLWR assumption to reduce the key
distinguishability under adaptive chosen ciphertext attacks (IND-CCA2)
                                                                                            sizes.
security. Due to the module variants of selected problems and the
efficiency of MLWR structure in encryption, SMAUG-T.KEM presents
reduced public key and ciphertext sizes and running time results.                           Definition 2 (MLWR Problem [36]). Let {𝑝, 𝑞, 𝜂, 𝑘} ∈ Z+ such that
    The main procedures of SMAUG-T.KEM [11] are remembered in                               𝑞 ≥ 𝑝 ≥ 2, general public key 𝐴 ←𝑟 ℜ𝑘×𝑘                        𝑘
                                                                                                                                𝑞 , and secret key 𝑠 ←𝑟 𝑆𝜂 .
Table 3.                                                                                    MLWR distribution is defined by (𝐴, 𝑏 = ⌊ 𝑞𝑝 𝐴𝑠⌉) ∈ ℜ𝑘×𝑘
                                                                                                                                                 𝑞   × ℜ𝑘×1
                                                                                                                                                        𝑝 .


                                                                                        4
K. Seyhan et al.                                                                                                             Computer Standards & Interfaces 97 (2026) 104118


   The hardness of MLWR is determined by the Adv of 𝐀 to solve                        • new-session: It allows one of the communicating parties to ini-
decisional-MLWR:                                                                        tiate a connection with the other party utilizing a shared pass-
                     |                                                                  word. The built connection and password of the first party are
AdvMLWR                                𝑘×𝑘
    𝑛,𝑞,𝑝,𝑘,𝜂 (𝐀) = ||Pr[𝐛 = 1|𝐴 ←𝑟 ℜ𝑞 ; 𝑏 ←𝑟 ℜ𝑞 ;
                                                    𝑘
                                                                                        transcribed by using this query in the functionality.
                                         𝐛 ← 𝐀(𝐴, 𝑏)]−
                                                                                      • test-pw: Online dictionary attacks are modeled with this query.
                                                 𝑝
        Pr[𝐛 = 1|𝐴 ←𝑟 ℜ𝑘×𝑘             𝜂 ; 𝑏 ← ⌊ 𝑞 𝐴𝑠⌉
                                       𝑘
                            𝑞  ; 𝑠 ← 𝑆                                                  When test-pw is queried, it also changes the appearance of ideal
                                   𝑝        |                                           functionality during the key exchange, as the behavior of the
                   𝐛 ← 𝐀(𝐴, 𝑏 = ⌊ 𝐴𝑠⌉)]| < 𝑛𝑒𝑔𝑙(𝑛)                                      next query is altered based on guessing the correct or incorrect
                                   𝑞        |
                                                                                        password.
    The proposed PAKE mainly aims to obtain an efficient PAKE solu-
tion, even if it has complex KEM structures, additional encryption, or                • new-key: This query, modeled as an interface, allows the con-
primitives. To achieve, the currently proposed generic PAKE construc-                   nected parties to be given session keys consistent with their
tion from a well-structured KEM named as OCAKE was selected.                            records if the fresh parties utilize the same password.

Definition 3 (OCAKE Construction [12]). OCAKE-based PAKE is one of                   The explicit authentication of the server is analyzed under the pass-
the KEM to PAKE solutions in the literature that was built by consider-           word authentication-based ideal functionality assumptions described in
ing the traditional one-way encrypted key exchange idea [38]. It stands           Algorithm 1.
out with its ability to provide explicit password-based authentication
with single encryption. In addition to capture security in the relaxed            Algorithm 1 Ideal Functionality Definitions of a PAKE that Provide
model with static corruptions, it is a model to build efficient KEM to            Explicit Server Authentication
PAKE designs.                                                                      procedure Session-Initialization
                                                                                      For (new-session, s-sid, Ŕ, pw, 𝐂𝑖 , 𝐒𝑗 ) in 𝐂𝑖 ;
   In the OCAKE construction, selected KEM needs to satisfy the fuzzi-
ness and anonymity properties that define the randomness of public                      • (new-session, s-sid, Ŕ, 𝐂𝑖 , 𝐒𝑗 ) is sent to 𝐀.
keys and encapsulation, respectively.                                                   • If this is the first or second new-session query and there is a record
                                                                                                             ̄ ⋅, ⋅) ∈ L, (s-sid, 𝐂𝑖 , 𝐒𝑗 , pw, fresh, Ŕ) is recorded to L.
                                                                                          (s-sid, 𝐂𝑖 , 𝐒𝑗 , pw,

Definition 4 (Anonymity of a KEM [12,32]). The anonymity of a KEM                  end procedure
scheme is defined by analyzing the randomness of ciphertext distri-                procedure Active-Attack
                                                                                                                                                                                   ̄
                                                                                      If a record (s-sid, 𝐂𝑖 , 𝐒𝑗 , pw, fresh, Ŕ) ∈ L exists when 𝐀 queries (test-pw, s-sid, 𝐂𝑖 , pw),
bution. If the ciphertext distribution obtained by the encapsulation               the following reactions are done.
function of the KEM scheme is computationally indistinguishable from
                                                                                        • If pw = pw,
                                                                                                    ̄ the record is marked as compromised. The answer is labeled as
the uniform ciphertext distribution, it satisfies the anonymity property.
                                                                                          correct-guess and returned to 𝐀.
                                                                                        • Otherwise, the record is marked as interrupted. The answer is tagged as
Definition 5 (Fuzziness of a KEM [12,32]). Fuzziness is specified as                      incorrect-guess and returned to 𝐀.
a measure of the properties of the public key distribution. More pre-              end procedure
cisely, if the distribution of public keys cannot be computationally               procedure Key-Generation
distinguished from the uniform distribution, the KEM is said to be                    When a query (new-key, s-sid, 𝐂𝑖 , ssk) is received from §, where ssk ∈ {shared-keys}.
fuzzy.                                                                                  • In order to be the first new-key query for 𝐂𝑖 , there must be a record of the form
                                                                                          (s-sid, 𝐂𝑖 , 𝐒𝑗 , pw, stt, Ŕ) for any stt.
2.2. Security model
                                                                                              – If Ŕ=𝐂,

   In the literature, there have been used two different models, Bellare–                            ∗ If stt=compromised, or one of the parties 𝐂𝑖 or 𝐒𝑗 is cor-
Pointcheval–Rogaway (BPR) [39] and UC [40] to analyze the security                                     rupted, and there are two records such as (s-sid, 𝐂𝑖 , 𝐒𝑗 , pw, 𝐂) and
                                                                                                       (s-sid, 𝐒𝑗 , 𝐂𝑖 , pw, 𝐒), then (s-sid, ssk) values are sent to 𝐂𝑖 .
of PAKE schemes. In the BPR model, game-based analysis was pre-                                      ∗ Else If stt=fresh,
sented, which includes the adversary’s protocol breaking capabilities.
With the UC model, there was a simulation-based approach where                                                · For pw=pw,                                          ̄ 𝐂′ ) and as ssk
                                                                                                                          ̄ if there is a record (s-sid, 𝐒𝑗 , 𝐂𝑖 , pw,             ̄
                                                                                                                has already been transferred to fresh 𝐒𝑗 , (s-sid, ssk)̄ values are
the security guarantee is more certain. Due to the selected PAKE                                                sent to 𝐂𝑖 .
construction, in this paper, UC model is used to analyze the security                                         · For pw≠ pw,         ̄ ← {0, 1}𝑘 is chosen and (s-sid, ssk)
                                                                                                                             ̄ an ssk    𝑟
                                                                                                                                                                              ̄ are
of proposed PAKE to present strictly better guarantees. Let us define                                           transferred to 𝐂𝑖 .
UC-related security terms [12,40].                                                                   ∗ Else If stt=fresh,
      ∏
    •    -𝐀-§-Ŕ-𝐂-𝐒: Protocol-Adversary-Simulation-Role component-                                            · For this s-sid, if there is no completed record for 𝐒𝑗 , nothing
      Client-Server                                                                                             is done.

    • 𝐅: Ideal functionality is considered as a honest trusted party that                            ∗ Else If stt=interrupted.
      unconditionally answers to queries.
                                                                                                              · (s-sid, error) values are sent to 𝐂𝑖 .
    • sid-stt: Session identifier-Status component
    • s-sid = (sid, s-sid): Unique sub-session identifier                                     – If Ŕ=𝐒,

    • L: A List for session record, where L = (s-sid, 𝐂𝑖 , 𝐒𝑗 , pw, stt, Ŕ)                          ∗ If stt=compromised, or one of the parties 𝐂𝑖 or 𝐒𝑗 is corrupted.
    • Real world: The execution of the protocol are run between parties                                       · (s-sid, ssk) values are sent to 𝐂𝑖 .
      in the presence of an 𝐀.
                                                                                                     ∗ Else If stt=fresh, or stt=interrupted,
    • Ideal world: Dummy actors and an ideal adversary/simulator in-
      teract only with an 𝐅 to determine the output of special function.                                      · An ssk̄                                           ̄
                                                                                                                          ←𝑟 {shared-keys} is chosen and (s-sid, ssk) are
                                                                                                                transferred to 𝐂𝑖 .
                                                         ∏
   The main aim of UC analysis is to imitate a protocol ( ) by utilizing
                                                                                   end procedure
ideal functionality (𝐅). If the emulation cannot distinguish the outputs           The record is updated as completed.
of protocol in the presence of feasible adversary interactions from the
outputs of dummy actors and simulator (§) in the interaction of ideal
               ∏
functionality,    is considered UC-emulation functionality.
   Three different queries describe the ideal functionality of a PAKE                 • The role component, Ŕ= {𝐂, 𝐒}, is added to the session-related
protocol [40].                                                                          records kept in L.

                                                                              5
K. Seyhan et al.                                                                                                      Computer Standards & Interfaces 97 (2026) 104118


    • On the same session s-sid, if the client queries new-key at a time                 After showing the anonymity and fuzziness of SMAUG-T.KEM, the
      when the server is not still querying new-key, nothing is done.                proposed OCAKE-based SMAUG-T.PAKE scheme is detailed in Fig. 1.
    • The client is cancelled regardless of the status if the parties, 𝐂𝑖                In the constructed PAKE, four sub-processes are performed on the
      and 𝐒𝑗 , do not share the same password.                                       client and server sides. In these processes, SMAUG-T.KEM procedures,
                                                                                     given in Table 3, symmetric encryption and decryption, and two dif-
   In the security analysis, it is proved that the proposed PAKE is a                ferent hash functions are used to obtain password-authenticated key
UC-emulation in password-based functionality model. When examining                   sharing with explicit authentication. Note that in the implementation,
the protocol’s security, static corruptions are considered in the erasure            two symmetric encryption techniques, AES and Ascon, are used to cap-
model during the simulation, where the simulator knows which sides                   ture the best efficiency. The step-by-step explanation of the proposed
                                                                                     PAKE can be summarized as follows.
are corrupted. Note that in the theoretical security analysis, password
security is not the main concern. In the PAKE protocols, passwords are                  • Process 𝐂0 : The public key and secret key are determined by using
assumed to be securely shared between the parties by adding some                          SMAUG-T.KEM key generation, defined in Table 3. s-sid∥𝑝𝑤𝐂
boundaries to the selection and usage of passwords and multi-factor                       concatenation and pk components are provided as the actual
authentication solutions.                                                                 inputs of symmetric encryption. Finally, password encrypted 𝑝𝑘′
                                                                                          is sent to the server.
                                                                                        • Process 𝐒0 : Firstly, server decrypts 𝑝𝑘′ to obtain actual 𝑝𝑘 by
3. Proposed SMAUG-T.PAKE
                                                                                          using s-sid and 𝑝𝑤𝐶 . Then, the recovered 𝑝𝑘 is used to obtain
                                                                                          ciphertext 𝑐𝑡 and capsulated key 𝐾 with the help of SMAUG-T’s
    The proposed protocol focused on how a KEM scheme containing                          encapsulation procedure. Ultimately, server generates a key, 𝐾̄ as
performance-efficient components can be converted into an efficient                       an authentication tag and sends (𝑐𝑡, 𝐾)  ̄ pairs to the client.
PAKE. For this purpose, SMAUG-T.KEM [11], whose security is defined                     • Process 𝐂1 : As soon as the client receives the parameters from the
under MLWE and MLWR assumptions, is chosen as the KEM scheme.                             server, it generates the key 𝐾 ′ by recovering the ciphertext with
To add password-based authentication, the OCAKE model [12], which                         SMAUG-T’s decapsulation procedure. Then, it generates the au-
allows efficient and explicit authentication with single encryption, is                   thentication               check           component              with
followed. This model requires the KEM on which it is based to provide                     𝐾̄ ′ = 𝐻1 (s-sid∥𝐂∥𝐒∥𝑝𝑤𝐂 ∥𝑝𝑘′ ∥𝑐𝑡∥𝐾 ′ ). Finally, client computes it is
anonymity and fuzziness properties. Before giving the main protocol                       shared key with the help of computed {s-sid, 𝐂, 𝐒, 𝑝𝑘′ , 𝑐𝑡, 𝐾̄ ′ , 𝐾 ′ }
flow, let us prove the anonymity and fuzziness of SMAUG-T.KEM in                          if 𝐾̄ ′ == 𝐾̄ is hold.
Lemma 1.                                                                                • Process 𝐒1 : Server generates its shared key by using
                                                                                                                  ̄ 𝐾}.
                                                                                          {s-sid, 𝐂, 𝐒, 𝑝𝑘′ , 𝑐𝑡, 𝐾,
Lemma 1.                                                                                The most fundamental components that affect the correctness of
   (i) We refer to [11] to check the detailed proofs of indistinguishability.        the proposed scheme are {𝐾 ′ , 𝐾} such that 𝐾 ′ = 𝐾. As seen in Fig. 1,
                                                                                     these components are generated by using SMAUG-T’s encapsulation and
       Smaug.KEM on parameters (𝑛, 𝑝, 𝑞, 𝑘, 𝜂, ℎ𝑠, ℎ𝑟) holds this property if
                                                                                     decapsulation procedures that were remembered in Table 3. In [11],
                                                                                     the decryption failure of SMAUG-T.KEM is defined with 𝛿 = Pr[‖𝑟𝑇 ⋅
        Advind
           SMAUG-T
                   (𝐀) ≤AdvPRF
                           expandA
                                   (𝑡) + AdvMLWE
                                            𝑛,𝑞,𝑘,𝑘,𝜂,ℎ𝑠 (𝑡)+                        𝑒 + 𝑠𝑇 ⋅ 𝑒1 + 𝑒2 ‖∞ < 2𝑡𝑞 ] probability. The analysis showed that if the
                                                                          (1)        parameter set is selected by considering this condition, SMAUG-T.KEM
                          AdvMLWR
                             𝑛,𝑞,𝑝,𝑘+1,𝑘,𝜂,ℎ𝑟 (𝑡)
                                                                                     and SMAUG-T.PAKE will be run correctly with less than 𝛿 probability.
   (ii) To adapt OCAKE model into SMAUG-T.KEM, we proved that
                                                                                     4. Security analysis
        SMAUG-T.KEM satisfies the anonymity property in ℜ𝑘𝑝 × ℜ𝑝 :

        Advano     (𝐀) ≤AdvPRF     (𝑡) + AdvMLWE                                         In constructing SMAUG-T.PAKE, the explicitly authenticated generic
           SMAUG-T         expandA          𝑛,𝑞,𝑘,𝑘,𝜂,ℎ𝑠 (𝑡)+
                                                                          (2)        OCAKE model is integrated into the SMAUG-T.KEM algorithm to gen-
                          AdvMLWR
                             𝑛,𝑝,𝑞,𝑘+1,𝑘,𝜂,ℎ𝑟 (𝑡)                                    erate an efficient password-authenticated version. OCAKE structure
                                                                                     assumes the adversary can obtain the party’s password before the exe-
        Proof. Let a public key sample of SMAUG-T, (𝐴𝑇 |𝑏𝑇 )𝑇 ←𝑟 ℜ(𝑘+1)×𝑘
                                                                       𝑞  ,          cution. So, the simulator is aware of which party is broken or corrupted.
        is given. By rewriting the ciphertext 𝑐 = (𝑐1𝑇 , 𝑐2 ) ∈ ℜ𝑘+1
                                                                 𝑝   ,               The semantic security is ensured in the UC model with static corrup-
                        𝑀𝐿𝑊 𝑅                                                        tions if the underlying KEM provides fuzziness, indistinguishability, and
                 ⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞⏞                                                     anonymity.
           [ ]          [ ]        [ ]
            𝑐1    𝑝      𝐴       𝑝 0                                                     The security of the SMAUG-T.PAKE model is analyzed under the
        𝑐=     = ⌊ ⋅ 𝑇 ⋅ 𝑟⌉ + ⋅                                           (3)
            𝑐2    𝑞 𝑏            𝑡 𝜇                                                 password authenticated-based ideal functionality assumptions, defined
                                                                                     in Algorithm 1 by making adaptations to MLWE+MLWR assumptions.
        is obtained, where 𝑡|𝑝|𝑞. As long as the hardness of MLWR is
                                                                                     The advantage of an adversary against OCAKE-based SMAUG-T.PAKE
        satisfied, the distribution of 𝑐 is computationally indistinguishable
                                                                        [ ]          is analyzed in Theorem 1.
                                                                    𝑝    0
        from uniformly random sample in ℜ𝑘+1  𝑝   since the added   𝑡
                                                                      ⋅     is
                                                                         𝜇
                                    𝑘+1                                              Theorem 1. Let (Enc, Dec) and (𝐻1 , 𝐻2 ) be ideal ciphers and random
        also a random vector in ℜ𝑝 . □
                                                                                     oracle pairs, respectively. Let 𝑚𝐸 and 𝑚𝐷 be the maximum query numbers
  (iii) Like anonymity, we also show that SMAUG-T.KEM provides fuzzi-                for encryption (Enc) and decryption (Dec) oracles and 𝑚𝑆 as the number
        ness in ℜ𝑘𝑞 :                                                                of sessions. The semantic security of the proposed SMAUG-T.PAKE in the
                                                                                     UC model is specified by Eq. (5), based on the fuzziness, anonymity, and
             fuz
        AdvSMAUG-T (𝐀) ≤ AdvMLWE
                            𝑛,𝑞,𝑘,𝑘,𝜂,ℎ𝑠 (𝑡)
                                                                          (4)        indistinguishability properties of underlying KEM.
        Proof. As stated in Table 3, the public key of SMAUG-T.KEM is                AdvSMAUG-T.KEM
                                                                                        SMAUG-T.PAKE
                                                                                                     (𝐀) ≤ AdvSMAUG-T.KEM
                                                                                                              fuzz
                                                                                                                          (𝑡) ⋅ (𝑚𝐷 + 𝑚𝑆 )+
        generated with SMAUG-T.PKE.KeyGen algorithm. Since pk = (𝑝, 𝑏),                                 AdvSMAUG-T.KEM (𝑡) ⋅ 𝑚𝐷 +
                                                                                                           ano
        where 𝑏 = −𝐴𝑇 𝑠 + 𝑒 mod 𝑞 and (𝑝, 𝜏) ←XOF(seed) is generated by
                                                                                                        AdvSMAUG-T.KEM (𝑡) ⋅ (𝑚𝑆 + 𝑚𝐷 + 1)+                       (5)
        following MLWE assumption, the distribution of pk is computation-                                  ind

        ally indistinguishable from uniformly random sample in {0, 1}256 ×                              (𝑚𝐻1 + 𝑚𝐻2 ) ⋅ 𝑚𝑆 ⋅ 2−𝑛 +
        ℜ𝑘𝑞 . □                                                                                         𝑚2𝐸 ⋅ 2−𝜅 + 𝑚𝐻1 ⋅ 2−𝑛

                                                                                 6
K. Seyhan et al.                                                                                               Computer Standards & Interfaces 97 (2026) 104118




                                                  Fig. 1. Proposed OCAKE-based SMAUG-T.PAKE scheme.


                                                                                    achieved in either scenario, the probability of environment in
Proof. For the sketch of proof, the simulated game sequence Game𝑖                   Game5 is defined by Eq. (10).
is defined, where 𝑖 = {0, … , 8}. The game series starts with real
                                                                                    |Pr[Game5 ] − Pr[Game4 ]| = 0                                        (10)
game Game0 and ends with ideal game Game8 that uses only its ideal
functionality defined in Algorithm 1.                                              • Game6 ∶ In Game6 , client’s second reaction is simulated. In the
                                                                                     simulation, § simulates the answer of honest client when it gets
    • Game0 ∶ Game0 that defines the real-world protocol is identified                    ̄ Since these components can come either from an honest
                                                                                     (𝑐𝑡, 𝐾).
      by considering indistinguishable, anonymous, and fuzzy KEM un-                 server or from an adversary who corrupts the server, §’s behavior
      der the random oracle, erasure, ideal cipher, and static corruption            is determined accordingly. In the first case, there is no difference
      assumptions.                                                                   from Game5 since the honest version is evaluated. In the second
    • Game1 ∶ The ideal cipher and two random oracle simulations are                 scenario, the cancellation situation will occur due to the 𝐾̄ ′ ≠ 𝐾.
                                                                                                                                                       ̄
      modeled with this game. Let 𝑃 𝑟[Game1 ] be the probability of the              So, there will be no difference from Game5 , and the probability
      environment for outputting 1 in the simulated Game1 . Under the                of environment in Game6 is presented in Eq. (11).
      assumptions of random oracles 𝐻1 and 𝐻2 and ideal cipher Enc,
                                                                                    |Pr[Game6 ] − Pr[Game5 ]| = 0                                        (11)
      given in Algorithm 1, the environment can differentiate the real
      protocol execution from Game1 when § aborts. The probability of              • Game7 ∶ In Game7 , ciphertext (𝑐𝑡), authentication tag compo-
      environment in Game1 is given in Eq. (6).                                      nents (𝐾, 𝐾 ′ ), and shared keys ({ssk𝐂 , ssk𝐒 }) are replaced with
                                                                                     random values. Three situations arise.
       |Pr[Game1 ] − Pr[Game0 ]| ≤ 𝑚2𝐸 ⋅ 2−𝜅−1 + 𝑚2𝐻 ⋅ 2−𝑛−1            (6)
                                                    1
                                                                                        – Randomization of 𝑐𝑡:
    • Game2 ∶ The random secret keys are embedded during the
      simulation of decryption oracle Dec. So, the probability of en-                         ∗ By utilizing indistinguishability of the SMAUG-T.KEM,
      vironment in Game2 is associated with fuzziness property of                               AdvSMAUG-T.KEM
                                                                                                    ind
                                                                                                                (𝑡) ⋅ 𝑚𝐷 is defined the bound in the
                                                                                                server’s computations since 𝑝𝑘 comes from decryption
      underlying KEM, proved in Lemma 1, since the difference of real
                                                                                                Dec is obtained.
      and random public key is defined with the this property and given
                                                                                              ∗ On the client’s part, the indistinguishability of the
      in Eq. (7).
                                                                                                Smaug.KEM builds the bound without any other com-
       |Pr[Game2 ] − Pr[Game1 ]| ≤ AdvSMAUG-T.KEM (𝑡) ⋅ 𝑚𝐷              (7)                     ponents due to the randomization of 𝐾 ′ . So, the bound
                                      fuzz
                                                                                                is AdvSMAUG-T.KEM
                                                                                                       ind
                                                                                                                   (𝑡).
    • Game3 ∶ In Game3 , the adversary’s capacity to estimate 𝐾̄ with-
      out asking the correct query to 𝐻1 is modeled. If this case hap-                  – Randomization of authentication checks: The environment
      pens, the § will be cancelled. So, the probability of environment                   with random components can fail the game in two ways.
      in Game3 is defined by Eq. (8).                                                         ∗ Server’s side: 𝑐𝑡 is selected randomly from
       |Pr[Game3 ] − Pr[Game2 ]| ≤ 𝑚𝑆 ⋅ 2−𝑛
                                                                        (8)                     ciphertext distribution instead of (𝑐𝑡, 𝐾)           ←
                                                                                                SMAUG-T.KEM.EnCap(𝑝𝑘) computation. Since this
    • Game4 ∶ In Game4 , the client’s initialization is simulated by uti-                       case is also associated with anonymity, 𝐀 can distin-
      lizing Dec rather than Enc. In the simulation, § selects a random                         guish this simulation by querying 𝑚𝐷 times decryption
      𝑝𝑘′ ←𝑟 2|𝐸| , requests 𝑝𝑘 = Dec(s-sid∥𝑝𝑤𝐂 , 𝑝𝑘′ ), and forwards 𝑝𝑘′ to                    Dec to break the anonymity. So, AdvSMAUG-T.KEM
                                                                                                                                        ano        (𝑡) ⋅
      server. These changes do not reveal any difference from Game3 ,                           𝑚𝐷 .
      the probability of environment in Game4 is defined by Eq. (9).                          ∗ Client’s side: Since the simulator can try to query 𝐻1
                                                                                                and K was truly random in the previous game, 𝑚𝐻1 ⋅
       |Pr[Game4 ] − Pr[Game3 ]| = 0                                    (9)                     𝑚𝑆 ⋅ 2−𝑛 is obtained.
    • Game5 ∶ In Game5 , server’s answer, (𝑐𝑡, 𝐾),
                                                 ̄ is simulated. In the                 – Randomization of shared keys (ssk𝐂 , ssk𝐒 ←𝑟 {0, 1}𝜅 ): The
      simulation, § simulates the answer of the honest server when                        only way the environment can detect the difference is
      it gets 𝑝𝑘′ . Since 𝑝𝑘′ can come either from an honest client or                    through 𝐻2 oracle calls. Since 𝐾 is truly random in the
      an adversary who corrupts the client, §’s behavior is determined                    previous game and there are at most 𝑚𝑆 changes, 𝑚𝐻2 ⋅ 𝑚𝑆 ⋅
      accordingly. Since no difference from the previous game can be                      2−𝑛 is procured.

                                                                               7
K. Seyhan et al.                                                                                               Computer Standards & Interfaces 97 (2026) 104118


       Finally, the probability of environment in Game7 is defined by Eq.       SMAUG-T functions used in the reference SMAUG-T.PAKE and pre-
       (12).                                                                    sented the comparative results in Table 6. The codes are written in C
                                                                                and given in [41].
       |Pr[Game7 ] − Pr[Game6 ]| ≤AdvSMAUG-T.KEM
                                     ind
                                                 (𝑡) ⋅ (𝑚𝐷 + 1)+
                                                                                    Table 6 implies that in both encryption modes, the SMAUG-T.PAKE
                                    AdvSMAUG-T.KEM
                                       ano         (𝑡) ⋅ 𝑚𝐷 +       (12)        scheme gives better results in terms of CPU cycle and runtime than
                                    (𝑚𝐻1 + 𝑚𝐻2 ) ⋅ 𝑚𝑆 ⋅ 2−𝑛                     the generic Kyber.PAKE scheme since SMAUG-T.PAKE benefits from the
                                                                                efficient algebraic structure of SMAUG-T.KEM.
    • Game8 ∶ In Game8 , the ideal world is modeled by adding ideal                 We also compared the performance of proposed PAKE with the
      functionality assumptions. According to Algorithm 1, there are            lattice-based PAKE protocols in the literature and whose accessible
      two possible fresh session cases.                                         codes were found. In order for the comparison to be meaningful,
                                                                                one-stage PAKE protocols were chosen based on the same hard lat-
           – If honest parties use the same password, stt = success will
                                                                                tice problem. Since the main security of SMAUG-T.PAKE is based on
             allow the same session keys to be obtained.
                                                                                MLWE+MLWR problems, MLWR-based Saber.PAK.PAKE [25], MLWE-
           – When honest parties use the same password, abort in ideal
                                                                                based MLWE.PAK.PAKE [23], MLWE-based Kyber.PAK.PAKE [30] are
             functionality is achieved. A random session key is returned
                                                                                selected to determine the efficiency of PAKE construction models.
             due to the stt = fail situation.
                                                                                    The performance results of these schemes are obtained by using
                                                                                the provided C codes, [23,25,30] and run on the same computer. The
                                                                                performances of selected module-based PAKEs with lattice assumptions
Finally, the total bound of the environment is given in Eq. (13).               are evaluated in terms of running times and consumed CPU cycles and
|Pr[Game8 ] − Pr[Game0 ]| =                                                     are given in Table 7.
                                                                                    Table 7 shows that the proposed SMAUG-T.PAKE shows the best
                              |Pr[Game8 ] − Pr[Game7 ]|+
                                                                                results among other module-based schemes, even if it has additional
                              |Pr[Game7 ] − Pr[Game6 ]|+                        ideal cipher usage and KEM components. The reason for this is the
                              |Pr[Game6 ] − Pr[Game5 ]|+                        KEM selection, which consists of efficient arithmetic operations, and
                              |Pr[Game5 ] − Pr[Game4 ]|+                        the generic model in the PAKE design.

                              |Pr[Game4 ] − Pr[Game3 ]|+                        5.2. Case scenario: Reference implementation on mobile environment
                              |Pr[Game3 ] − Pr[Game2 ]|+
                                                                    (13)            The efficient implementation of SMAUG-T.PAKE shows that it can
                              |Pr[Game2 ] − Pr[Game1 ]|+
                                                                                be one of the best options for providing post-quantum secure PAKE
                              |Pr[Game1 ] − Pr[Game0 ]|
                                                                                for mobile environments. To analyze mobile compatibility, the JAVA
                           ≤AdvSMAUG-T.KEM
                               fuzz
                                           (𝑡) ⋅ (𝑚𝐷 + 𝑚𝑆 )+                    codes of SMAUG-T.PAKE is also generated [41]. A computer with
                              AdvSMAUG-T.KEM
                                 ano         (𝑡) ⋅ 𝑚𝐷 +                         32 GB RAM and a hexa-core AMD Ryzen 5 5500 processor running
                                                                                at 3.60 GHz is used as a server while Samsung Galaxy A51 (8 Cores)
                              AdvSMAUG-T.KEM
                                 ind
                                             (𝑡) ⋅ (𝑚𝑆 + 𝑚𝐷 + 1)+
                                                                                with 4x 1.7 GHz ARM Cortex-A53 397 co-processor with 2.3 GHz and
                              (𝑚𝐻1 + 𝑚𝐻2 ) ⋅ 𝑚𝑆 ⋅ 2−𝑛 +                         4x 2.3 GHz ARM Cortex-A73 main processor is used as mobile device.
                              𝑚2𝐸 ⋅ 2−𝜅 + 𝑚𝐻1 ⋅ 2−𝑛   □                         The mobile performance results are obtained in terms of running time,
                                                                                memory and CPU usages and presented in Table 8.
5. Implementation details and discussion                                            To compare the efficiency of these mobile-based one round PAKE
                                                                                schemes, [30] is also examined under the same operating conditions
    In this section, performance comparison results are presented with          and the results are presented in Table 9 and visualized in Fig. 2. In
reference and mobile implementations to show the effectiveness of the           Fig. 2, mobile device performance metrics such as CPU usage, energy
proposed PAKE.                                                                  consumption and memory usage are analyzed using the integrated
                                                                                Android Profiler in Android Studio. Each scheme is run in real-time
5.1. Performance analysis of the proposed PAKE                                  on a mobile device, and the collected performance data is visually
                                                                                recorded. These recordings are obtained through the graphical interface
    For the performance analysis of constructed generic PAKE, we                provided by Android Profiler, capturing the execution of each stage
optionally adapted Ascon or AES instead of external encryption to               over a specified time interval.
check the different performance options. By using reference C codes                 Table 9 and Fig. 2 show that for the mobile environment, the
of SMAUG-T.KEM and making adaptations to explicit generic PAKE                  AES version of SMAUG-T.PAKE has the best performance. In other
additions, the implementation of SMAUG-T.PAKE is written in C and               application results, it is known that the Ascon module comes to the
can be found in [41]. Performance results are obtained on a computer            fore in performance. This result is explained by the fact that AES
with 32 GB RAM and a hexa-core (6 Core) AMD Ryzen 5 5500 proces-                consists of mobile-optimized generic codes in the JAVA library. Ascon’s
sor running at 3.60 GHz. Processor cycles (median and average) and              JAVA codes were not optimized for mobile, resulting in lower perfor-
runtime results are determined by averaging 10,000 runs. For three              mance. As a result, the recommended SMAUG-T.PAKE is the PAKE that
different security levels, performance results are obtained according           provides the best efficient application results for the mobile world.
to the parameter set in Table 4. The reference SMAUG-T.PAKE im-
plementation results in terms of two-different encryption methods are           5.3. Discussion
presented in Table 5.
    Table 5 shows that, as expected, lightweight Ascon presents better              The main focus of this paper is the design of a new PQC PAKE
results for three different security levels when used as the encryption         protocol using a combination of different cryptographic principles, such
method. Therefore, it is recommended that SMAUG-T.PAKE at Ascon                 as KEM and encryption/decryption primitives. It is aimed to evaluate
mode can be suitable for resource-constrained devices, while in other           the post-quantum security and efficiency effects of different primitives
applications both AES and Ascon can be used.                                    in the post-quantum secure PAKE design. Contributions are made to the
    We also analyzed how much the KEM selected in the PAKE design               design of the post-quantum secure PAKE protocol, theoretical security
model affects the performance. We integrated Kyber instead of the               analysis, and practical performance results. In order to make a fair

                                                                            8
K. Seyhan et al.                                                                                                                                                                                                       Computer Standards & Interfaces 97 (2026) 104118


Table 4
Parameter set.
 Scheme                                               Saber.PAK.PAKE                                                 MLWE.PAK.PAKE                                                    Kyber.PAK.PAKE                                          SMAUG-T.PAKE
                                                      [25]                                                           [23]                                                             [30]
 Security                                             128               192                256                       116                177                239                        128              192                256                 128                192                256
 level
 Module                            k                  2                 3                  4                         2                  3                  4                          2                3                  4                   2                  3                  5
 dimension
 Lattice                           n                  256               256                256                       256                256                256                        256              256                256                 256                256                256
 dimension
                                   q                  8192              8192               8192                      7681               7681               7681                       3329             3329               3329                1024               2048               2048
 Module
                                   p                  1024              1024               1024                      x                  x                  x                          x                x                  x                   256                256                256
                                   𝜂                  10                8                  6                         13                 8                  6                          x                x                  x                   x                  x                  x
 Distribution
                                   𝜂1                 x                 x                  x                         x                  x                  x                          3                2                  2                   x                  x                  x
 parameter
                                   𝜂1                 x                 x                  x                         x                  x                  x                          2                2                  2                   x                  x                  x
 Failure                           𝛿                  2−120             2−136              2−165                     2−53.4             2−97.4             2−131.6                    2−131            2−164              2−174               2−120              2−136              2−167
 rate


Table 5
Performance results of SMAUG-T.PAKE.
 Security level             128                                                                                      192                                                                                     256
 Cipher option              ASCON                                           AES                                      ASCON                                          AES                                      ASCON                                    AES
 Metrics                    M                    A            ET            M             A              ET          M              A               ET              M             A           ET             M                A          ET           M              A             ET
                     𝐂0     67 571               68 345       17 981        95 975        97 996         26 083      115 703        110 159         31 360          164 879       168 750     45 128         207 971          209 150    58 048       283 031        284 934       78 111
                     𝐒0     70 991               71 489       18 857        117 359       117 824        31 730      110 159        110 602         29 717          199 223       200 136     52 988         218 879          220 093    61 119       350 675        352 776       96 950
 Phases
                     𝐂1     91 367               91 825       24 507        90 683        90 859         24 234      141 299        142 489         38 547          141 263       141 123     44 128         259 739          261 757    72 709       259 523        261 267       71 542
                     𝐒1     12 491               12 794       2554          12 491        12 484         2468        17 891         18 004          3996            17 891        17 917      3977           26 927           27 118     7533         26 891         27 048        6509
 Total 𝐂                    158 938              160 170      42 488        186 658       188 855        50 317      257 002        252 648         69 907          306 142       309 873     89 256         467 710          470 907    130 757      542 554        546 201       149 653
 Total 𝐒                    83 482               84 283       21 411        129 850       130 308        34 198      128 050        128 606         33 713          217 114       218 053     56 965         245 806          247 211    68 652       377 566        379 824       103 459
 Total                      242 420              244 453      63 899        316 508       319 163        84 515      385 052        381 254         103 620         523 256       527 926     146 221        713 516          718 118    199 409      920 120        926 025       253 112



Table 6
Generic Kyber.PAKE vs. Proposed generic SMAUG-T.PAKE.
 Security level                            128                                                                             192                                                                               256
 Ideal Cipher                              ASCON                                    AES                                    ASCON                                        AES                                  ASCON                                   AES
 Metrics                                   M              A         ET              M             A             ET         M                A            ET             M             A        ET            M                A          ET          M           A                 ET
                           Total 𝐂 299 374                300 007   83,336 299 590 300 541 83,484 460 618                                   466 529      129,592        466 314 467 893 129,961 669 562 670 785 186,329                              673 522 675 909 187,752
 Generic
                           Total 𝐒 180 610                180 929   50,259 189 682 190 498 52,916 268 846                                   272 226      75,619         285 370 286 199 79,500 384 874 385 520 107,089                               405 214 406 273 112,854
 Kyber.PAKE
                           Total   479 984                480 936   133,595 489 272 491 039 136,400 729 464                                 738 755      205,211        751 684 754 092 209,461 1054 436 1056 305 293,418                            1078 736 1082 182 300,606
                           Total 𝐂 158 938                160 170   42,488          186 658 188 855 50,317                 257 002          252 648      69,907         306 142 309 873 89,256               467 710          470 907    130,757     542 554     546 201           149,653
 SMAUG-T.PAKE              Total 𝐒 83 482                 84 283    21,411          129 850 130 308 34,198                 128 050          128 606      33,713         217 114 218 053 56,965               245 806          247 211    68,652      377 566     379 824           103,459
                           Total           242 420 244 453 63,899                   316 508 319 163 84,515                 385 052 381 254 103,620 523 256 527 926 146,221 713 516                                            718 118    199,409 920 120         926 025           253,112



Table 7
A performance comparison for module-based PAKE schemes.
                                                                                                                                                                        Generic Kyber.PAKE                                                        SMAUG-T.PAKE
                            Saber.PAK.PAKE                          MLWE.PAK.PAKE                             Kyber.PAK.PAKE
Security                          [25]                                  [23]                                        [30]                                      AES                              ASCON                                    AES                              ASCON
           Metrics
 Level
                       𝐂               𝐒             Total      𝐂               𝐒         Total         𝐂            𝐒          Total           𝐂             𝐒          Total            𝐂        𝐒           Total          𝐂          𝐒        Total      𝐂              𝐒           Total

             M       186 226    116 314           302 540     207 784       217 561     425 345       314 566   172 690       487 256       299 590      189 682        489 272    299 374    180 610      479 984       186 658    129 850   316 508      158 938       83 482    242 420
  128        A       188 351    117 389           305 740     208 697       218 861     427 558       315 632   173 564       489 196       300 541      190 498        491 039    300 007    180 929      480 936       188 855    130 308   319 163      160 170       84 283    244 453
             ET       52,320     32,559            84,879      62,549        61,921     124,470        87,677    48,213       135,890        83,484       52,916        136,400     83,336     50,259      133,595        50,317     34,198    84,515       42,488       21,411     63,899

             M       309 652    184 678           494 330     319 104       318 672     637 776       498 994   271 654       770 648       466 314      285 370        751 684    460 618    268 846      729 464       306 142    217 114   523 256      257 002       128 050   385 052
  192        A       310 952    185 318           496 270     319 321       318 349     637 670       504 388   273 281       777 669       467 893      286 199        754 092    466 529    272 226      738 755       309 873    218 053   527 926      252 648       128 606   381 254
             ET       86,376     51,478           137,854      84,230        84,927     169,157       140,130    75,912       216,042       129,961       79,500        209,461    129,592     75,619      205,211        89,256     56,965   146,221       69,907        33,713   103,620

             M       465 478    267 838           733 316     449 640       428 235     877 875       696 490   373 030       1069 520      673 522      405 214     1078 736      669 562    384 874      1054 436      542 554    377 566   920 120      467 710       245 806   713 516
  256        A       469 478    270 625           740 103     445 296       422 208     867 504       699 656   374 171       1073 827      675 909      406 273     1082 182      670 785    385 520      1056 305      546 201    379 824   926 025      470 907       247 211   718 118
             ET      130,411     75,174           205,585     119,605       117,116     236,721       194,349   103,937        298,286      187,752      112,854      300,606       186,329    107,089       293,418     149,653    103,459   253,112      130,757        68,652   199,409




comparison, PAKE protocols defined on the same algebraic structure,                                                                                      would be ideal to compare protocols using the same design idea and
module, were selected even though they contain different design ideas.                                                                                   methodology, each protocol often differs from the other due to the
    To highlight the performance differences, firstly, the properties of                                                                                 targeted additional features. For example, the compromise structure,
selected module-based PAKEs are presented in Table 10. The main                                                                                          the difficulty problem, and additional principles used lead to different
                                                                                                                                                         evaluations. In this paper, we contribute to the literature by defining
characteristics of those schemes that reveal the differences regarding
                                                                                                                                                         security analysis in the UC model using hybrid hardness assumptions
performances appear to be due to the design model, main security,
                                                                                                                                                         for KEM to PAKE design methods, and by evaluating performance
reconciliation idea, and additional primitives such as ideal cipher and
                                                                                                                                                         in different encryption modes. We present in-depth security analysis
KEM usage.                                                                                                                                               evaluations by updating the defined methods under different security
    As summarized in Table 10, the PAKE protocol design incorporates                                                                                     assumptions. We also evaluate usability and applicability by imple-
different components and design ideas. This leads to different scenar-                                                                                   menting the results in different modes on different platforms. We
ios in both security analysis and performance evaluations. While it                                                                                      present theoretical and practical analyses on whether security and

                                                                                                                                                    9
K. Seyhan et al.                                                                                                               Computer Standards & Interfaces 97 (2026) 104118




Table 8
Mobile implementation results of SMAUG-T.PAKE.
 Security          Used             AES                                                                ASCON
 levels            encryption
                   Metrics          𝐂0        𝐒0          𝐂1        𝐒1       Total 𝐂       Total 𝐒     𝐂0        𝐒0            𝐂1          𝐒1         Total 𝐂      Total 𝐒
                   ET               640,772   611,31      491,996   76,133   1132,768 687,443          994,023   1032,032 527,931          82,921     1521,954 1114,953
 128               MU               58,2      47,2        57,4      3,5      115,6    50,7             337,2     344,4    57,3             3,5        394,5    347,9
                   CPU              8%        6%          8%        2%       16%      8%               12%       10%      8%               2%         20%      12%
                   ET               826,574   766,036     710,122   105,035 1536,696 871,071           1495,424 1426,737 732,831           107,363 2228,255 1534,1
 192               MU               93,1      81,5        96,5      4,9     189,6    86,4              481,8    408,6    96,7              4,9     578,5    413,5
                   CPU              9%        8%          10%       3%      19%      11%               16%      14%      11%               3%      27%      17%
                   ET               1111,587 1100,146 1115,541 146,240 2227,128 1246,386 2326,828 2363,742 1225,591 157,287 3552,419 2521,029
 256               MU               185,1    174,5    198,3    7,2     383,4    181,7    840,1    690,1    198,5    7,2     1038,6   697,3
                   CPU              12%      10%      13%      4%      25%      14%      25%      21%      12%      4%      37%      25%
–ET: Elapsed time in microsecond –MU: Memory usage in kilobayt –CPU: CPU usage
–The source codes is given in [41].


Table 9
A comparison for lattice-based PAKE schemes for mobile environment.
 Security             Used                Kyber.PAK.PAKE                             SMAUG-T.PAKE
 levels               encryption          [30]
                                          x                                          AES                                             ASCON
                      Metrics             Total 𝐂        Total 𝐒     Total           Total 𝐂         Total 𝐒     Total               Total 𝐂        Total 𝐒        Total
                      ET                  1645,489       1249,074    2894,563        1088,904        680,411     1768,508            1448,841       1089,582       2538,423
 128                  MU                  274,5          90,2        364,7           113,1           45,8        158,9               264,3          89,1           353,4
                      CPU                 16%            17%         33%             14%             8%          22%                 16%            17%            33%
                      ET                  2015,363       1498,580    3513,943        1517,179        879,821     2397,000            2193,568       1591,124       3784,692
 192                  MU                  361,6          135,3       496,9           186,1           84,4        270,5               359,1          134,4          493,5
                      CPU                 22%            19%         41%             17%             10%         27%                 22%            19%            41%
                      ET                  2825,102       1965,160    4790,262        2234,221        1282,773    3516,994            2825,102       1965,16        4790,262
 256                  MU                  477,1          173,8       650,9           377,9           178,6       556,5               477,1          173,8          650,9
                      CPU                 25%            23%         48%             22%             12%         35%                 25%            23%            48%


Table 10
The basic characteristics of module-based PAKEs.
                                Saber.PAK.PAKE              MLWE.PAK.PAKE              Kyber.PAK.PAKE                 Generic                       Proposed generic
                                [25]                        [23]                       [30]                           Kyber.PAKE+                   Smaug.PAKE
 Construction                   Traditional PAK PAKE        Traditional PAK PAKE       Traditional PAK PAKE           Generic PAKE from             Generic PAKE from
 model                          [1] from                    [1]                        [1] from                       KEM with the usage of         KEM with the usage of
                                well-structured KE                                     well-structured KEM            ideal cipher [12]             ideal cipher [12]
 Password                       Password is added as a      Password is added as a     Password is added as a         The password is used          The password is used
 usage                          component of public         component of public        component of public            as a parameter to             as a parameter to
                                key to provide              key to provide             key to provide                 encrypt public key and        encrypt public key and
                                authentication.             authentication.            authentication.                generate the shared           generate the shared
                                                                                                                      key component to help         key component to help
                                                                                                                      authentication.               authentication.
 Main                           MLWR                        MLWE                       MLWE                           MLWE                          MLWE+MLWR
 security
 Reconciliation                 bits                        OKCN                       Compress-Decompress            Compress-Decompress           Rounding function
 Additional                     X                           X                          KEM                            KEM                           KEM
 Primitives                                                                                                           Ideal Cipher                  Ideal Cipher
 Ideal                          X                           X                          X                              AES                           AES
 Cipher                                                                                                               ASCON                         ASCON
 Additional                     X                           X                          X                              Anonymity                     Anonymity
 requirements                                                                                                         fuzziness                     fuzziness
 Number of Hash                 4                           4                          3 + 3∗                         2 + 3∗                        2 + 3∗
 Security                       ROM                         ROM                        ROM                            UC                            UC
 model
*: The number of hash functions that were used as a component of KEM.+ : The implementation of generic Kyber PAKE is written in C to make a comparison.
OKCN: Optimally-balanced key consensus with noise.




                                                                                      10
K. Seyhan et al.                                                                                              Computer Standards & Interfaces 97 (2026) 104118


                                                                              performance in the lightweight encryption mode, even if it includes
                                                                              additional encryption and hash functions.
                                                                                  To demonstrate the efficiency of the proposed model, we also
                                                                              examined the performance of generic PAKE using Kyber.KEM instead of
                                                                              SMAUG-T.KEM. The results in Table 7 show that SMAUG-T is more ef-
                                                                              ficient than Kyber even when converted to PAKE. The security analysis
                                                                              presented in Section 4 demonstrates that the post-quantum security is
                                                                              maintained in the situation of KEM-to-PAKE transformation. The results
                                                                              presented in Table 9 also provide an analysis of the usability of lattice-
                                                                              based PAKE protocols on the mobile platform. These analyses show
                                                                              that the proposed PAKE can provide good performance in the standard
                                                                              algorithm mode even if it includes additional cryptographic primitives.
                                                                              While it is normally expected to provide more efficient results with
                                                                              lightweight algorithms such as Ascon, it has been evaluated that the
                                                                              reason why it provides more efficient results with AES is due to AES’s
                                                                              optimizations in the library functions used in the application.
                                                                                  This paper primarily presents evidence for the properties of
                                                                              anonymity and fuzziness associated with the conversion of a crypto-
                                                                              graphic principle designed as an efficient KEM for the post-quantum
                                                                              world to PAKE. It then analyzes the post-quantum security with hybrid
                                                                              security definitions created in the UC model, where the PAKE trans-
                                                                              formation retains its post-quantum security. Subsequently, we present
                                                                              usability and applicability analyses, comparing application results with
                                                                              literature solutions on various platforms. In addition to the theoretical
                                                                              and practical focus, efficient hardware-based implementations can also
                                                                              be implemented. For example, hardware-based analyses and imple-
                                                                              mentations for efficient implementations of PQC KEM algorithms have
                                                                              been carried out in studies such as [10,42–45]. Therefore, investigating
                                                                              more efficient versions of the SMAUG-T.PAKE protocol using different
                                                                              architectures and PUF-like hardware-based solutions will be conducted
                                                                              as future work.

                                                                              6. Conclusion

                                                                                  The design of post-quantum secure key-sharing schemes is one
                                                                              of the challenging issues in the literature. The simple structured au-
                                                                              thentication idea of PAKE schemes in real-world scenarios has also
                                                                              revealed the need to ensure post-quantum security of PAKE protocols.
                                                                              Therefore, PAKE schemes, which stand out with their strong security
                                                                              and efficiency features, are one of the necessary primitives for the post-
                                                                              quantum security of resource-limited devices. In this paper, we con-
                                                                              struct an efficient PAKE adaptation from well-structured lattice-based
                                                                              KEM procedures and additional primitives. The constructed SMAUG-
                                                                              T.PAKE, benefits the underlying KEM’s efficiency and simple structured
                                                                              KEM to PAKE construction. It is the first MLWE+MLWR-based KEM
                                                                              to PAKE design that provides explicit password-based anonymous and
                                                                              fuzziness authentication. Unlike the lattice-based PAKE protocols in the
                                                                              literature, the security analysis is performed under hybrid assumptions.
Fig. 2. Energy, CPU, and memory usage comparison diagrams for mobile          MLWE+MLWR-based password-authenticated key ideal functionality
compatible PAKEs.                                                             under the UC model are constructed to analyze the security of the
                                                                              proposed PAKE. The detailed performance analysis with other module-
                                                                              based, KEM-based, and mobile-based PAKE schemes shows that the
performance objectives can be maintained while creating new crypto-           proposed PAKE provides the best results in terms of consumed CPU
graphic primitives from known efficient ones, with the design of the          cycles and elapsed times, even if it contains additional encryption
post-quantum secure PAKE protocol.                                            usage. To the best of our knowledge, the proposed SMAUG-T.PAKE is
   A comparison is made with protocols based on the hardness as-              one of the best candidates for efficient password-based authentication
sumption of the MLWE and MLWR problem defined on the module                   for the post-quantum security of general purposes and mobile usage.
algebraic structure. The performance analyses for five different PAKE
protocols, based on the PAK-PAKE [1] and KEM-to-PAKE [12] design              CRediT authorship contribution statement
ideas, are presented in Table 7. Note that the performance evaluations
were conducted only with lattice-based PAKE protocols for which the              Kübra Seyhan: Writing – review & editing, Writing – original
source code is available, since the source code for other KEM to              draft, Validation, Methodology, Investigation, Conceptualization. Sedat
PAKE frameworks [12,31–34] in the literature is not shared. For fair          Akleylek: Writing – review & editing, Validation, Supervision, Project
evaluation, all code was obtained by running on the platforms. The            administration, Methodology. Ahmet Faruk Dursun: Writing – review
results show that, unlike [23,25,30], SMAUG-T.PAKE offers the best            & editing, Software.

                                                                         11
K. Seyhan et al.                                                                                                                 Computer Standards & Interfaces 97 (2026) 104118


Declaration of competing interest                                                           [17] J. Zhang, Y. Yu, Two-round PAKE from approximate SPH and instantiations from
                                                                                                 lattices, in: International Conference on the Theory and Application of Cryptology
                                                                                                 and Information Security, Springer, 2017, pp. 37–67, http://dx.doi.org/10.1007/
    The authors declare the following financial interests/personal rela-                         978-3-319-70700-6_2.
tionships which may be considered as potential competing interests:                         [18] Z. Zhao, S. Ma, P. Qin, Password authentication key exchange based on key
Sedat Akleylek reports financial support was provided by Estonian                                consensus for iot security, Clust. Comput. 26 (1) (2023) 1–12, http://dx.doi.
Research Council. If there are other authors, they declare that they have                        org/10.1007/s10586-022-03665-5.
                                                                                            [19] L. Chen, T. Qu, A. Yin, Quantum-safe multi-server password-based authenticated
no known competing financial interests or personal relationships that
                                                                                                 key exchange protocol, Multimedia Tools Appl. 83 (24) (2024) 65011–65038,
could have appeared to influence the work reported in this paper.                                http://dx.doi.org/10.1007/s11042-023-17984-1.
                                                                                            [20] Z. Li, D. Wang, E. Morais, Quantum-safe round-optimal password authentication
Acknowledgments                                                                                  for mobile devices, IEEE Trans. Dependable Secur. Comput. 19 (3) (2020)
                                                                                                 1885–1899, http://dx.doi.org/10.1109/TDSC.2020.3040776.
                                                                                            [21] A. Singh, H. Chandra, S. Rana, A robust lattice-based post-quantum three-party
    The authors would like to express their gratitude to the anonymous                           key exchange scheme for mobile devices, Concurr. Comput.: Pr. Exp. 37 (6–8)
reviewers for their invaluable suggestions in putting the present study                          (2025) e70036, http://dx.doi.org/10.1002/cpe.70036.
into its final form. Sedat Akleylek was supported by the Estonian                           [22] D. Mishra, K. Pursharthi, M. Singh, A. Mishra, Construction of post quantum
                                                                                                 secure authenticated key agreement protocol for dew-assisted IoT systems, Int.
Research Council Grant PRG2531 and Estonian Ministry of Defence
                                                                                                 J. Inf. Secur. 24 (1) (2025) 19, http://dx.doi.org/10.1007/s10207-024-00932-x.
Grant 2-2/24/541-1.                                                                         [23] P. Ren, X. Gu, Z. Wang, Efficient module learning with errors-based post-
                                                                                                 quantum password-authenticated key exchange, IET Inf. Secur. 17 (1) (2023)
Data availability                                                                                3–17, http://dx.doi.org/10.1049/ise2.12094.
                                                                                            [24] R. Ding, C. Cheng, Y. Qin, Further analysis and improvements of a lattice-
                                                                                                 based anonymous pake scheme, IEEE Syst. J. 16 (3) (2022) 5035–5043, http:
    No data was used for the research described in the article.                                  //dx.doi.org/10.1109/JSYST.2022.3161264.
                                                                                            [25] K. Seyhan, S. Akleylek, A new password-authenticated module learning with
                                                                                                 rounding-based key exchange protocol: Saber. PAKE, J. Supercomput. 79 (16)
References                                                                                       (2023) 17859–17896, http://dx.doi.org/10.1007/s11227-023-05251-x.
                                                                                            [26] C. Liu, Z. Zheng, K. Jia, Q. You, Provably secure three-party password-based
                                                                                                 authenticated key exchange from RLWE, in: International Conference on In-
 [1] P. MacKenzie, The PAK suite: Protocols for password-authenticated key exchange,
                                                                                                 formation Security Practice and Experience, Springer, 2019, pp. 56–72, http:
     Contrib. To IEEE P 1363 (2) (2002).
                                                                                                 //dx.doi.org/10.1007/978-3-030-34339-2_4.
 [2] F. Hao, P.C. van Oorschot, Sok: password-authenticated key exchange–theory,
                                                                                            [27] V. Dabra, A. Bala, S. Kumari, LBA-PAKE: Lattice-based anonymous password
     practice, standardization and real-world lessons, in: Proceedings of the 2022
                                                                                                 authenticated key exchange for mobile devices, IEEE Syst. J. 15 (4) (2020)
     ACM on Asia Conference on Computer and Communications Security, 2022, pp.
                                                                                                 5067–5077, http://dx.doi.org/10.1109/JSYST.2020.3023808.
     697–711, http://dx.doi.org/10.1145/3488932.3523256.
                                                                                            [28] V. Dabra, S. Kumari, A. Bala, S. Yadav, SL3PAKE: simple lattice-based three-party
 [3] J. Jiang, D. Wang, Qpase: Quantum-resistant password-authenticated searchable
                                                                                                 password authenticated key exchange for post-quantum world, J. Inf. Secur. Appl.
     encryption for cloud storage, IEEE Trans. Inf. Forensics Secur. 19 (2024)
                                                                                                 84 (2024) 103826, http://dx.doi.org/10.1016/j.jisa.2024.103826.
     4231–4246.
                                                                                            [29] S. Guo, Y. Song, S. Guo, Y. Yang, S. Song, Three-party password authentication
 [4] S.M. Bellovin, M. Merritt, Encrypted key exchange: Password-based protocols
                                                                                                 and key exchange protocol based on mlwe, Symmetry 15 (9) (2023) 1750,
     secure against dictionary attacks, 1992, http://dx.doi.org/10.7916/D8833ZSK.
                                                                                                 http://dx.doi.org/10.3390/sym15091750.
 [5] F. Hao, Prudent practices in security standardization, IEEE Commun. Stand. Mag.
                                                                                            [30] K. Seyhan, S. Akleylek, A.F. Dursun, Password authenticated key exchange-
     5 (3) (2021) 40–47, http://dx.doi.org/10.1109/MCOMSTD.121.2100005.
                                                                                                 based on kyber for mobile devices, PeerJ Comput. Sci. 10 (2024) e1960, http:
 [6] K. Seyhan, S. Akleylek, A comprehensive comparison of lattice-based password
                                                                                                 //dx.doi.org/10.7717/peerj-cs.1960.
     authenticated key exchange protocols defined on modules, in: International
                                                                                            [31] A. Arriaga, M. Barbosa, S. Jarecki, M. Škrobot, C’est très CHIC: A compact
     Conference on Information Technologies and their Applications, Springer, 2024,
                                                                                                 password-authenticated key exchange from lattice-based KEM, in: International
     pp. 91–105, http://dx.doi.org/10.1007/978-3-031-73417-5_8.
                                                                                                 Conference on the Theory and Application of Cryptology and Information
 [7] P.W. Shor, Algorithms for quantum computation: discrete logarithms and fac-                 Security, Springer, 2024, pp. 3–33, http://dx.doi.org/10.1007/978-981-96-0935-
     toring, in: Proceedings 35th Annual Symposium on Foundations of Computer                    2_1.
     Science, IEEE, 1994, pp. 124–134, http://dx.doi.org/10.1109/SFCS.1994.365700.          [32] J. Pan, R. Zeng, A generic construction of tightly secure password-based authen-
 [8] National Institute of Standards and Technology (NIST), NIST post-quantum                    ticated key exchange, in: International Conference on the Theory and Application
     cryptography standardization project, 2025, (Accessed: 24 October 2025) https:              of Cryptology and Information Security, Springer, 2023, pp. 143–175, http:
     //csrc.nist.gov/projects/post-quantum-cryptography.                                         //dx.doi.org/10.1007/978-981-99-8742-9_5.
 [9] D. Ott, C. Peikert, et al., Identifying research challenges in post quantum            [33] N. Alnahawi, J. Alperin-Sheriff, D. Apon, G.T. Davies, A. Wiesmaier, NICE-
     cryptography migration and cryptographic agility, 2019, https://arxiv.org/abs/              PAKE: On the security of KEM-based PAKE constructions without ideal ciphers,
     1909.07353.                                                                                 2024, URL https://eprint.iacr.org/2024/1957 Cryptology ePrint Archive, Paper
[10] N. Alnahawi, D. Haas, E. Mauß, A. Wiesmaier, SoK: PQC PAKEs - design, security              2024/1957.
     and performance, 2025, URL https://eprint.iacr.org/2025/119 Cryptology ePrint          [34] J. Vos, S. Jarecki, C.A. Wood, C. Yun, S. Myers, Y. Sierra, A hybrid asymmetric
     Archive, Paper 2025/119.                                                                    password-authenticated key exchange in the random oracle model, 2025, URL
[11] J. Cheon, H. Choe, J. Choi, D. Hong, J. Hong, C. Jung, H. Kang, J. Lee, S.                  https://eprint.iacr.org/2025/1343 Cryptology ePrint Archive, Paper 2025/1343.
     Lim, A. Park, S. Park, J. Seo, H. Seong, J. Shin, SMAUG-T: The Key Exchange            [35] J. Ding, S. Alsayigh, J. Lancrenon, S. Rv, M. Snook, Provably secure password
     Algorithm Based on Module-LWE and Module-LWR, Algorithm specifications,                     authenticated key exchange based on RLWE for the post-quantum world, in:
     K-PQC Consortium, 2024, (Accessed: 24 October 2025).                                        Cryptographers’ Track At the RSA Conference, Springer, 2017, pp. 183–204,
[12] H. Beguinet, C. Chevalier, D. Pointcheval, T. Ricosset, M. Rossi, GeT a CAKE:               http://dx.doi.org/10.1007/978-3-319-52153-4_11.
     Generic transformations from key encaspulation mechanisms to password authen-          [36] J.-P. D’Anvers, A. Karmakar, S. Sinha Roy, F. Vercauteren, Saber: Module-
     ticated key exchanges, in: International Conference on Applied Cryptography and             LWR based key exchange, CPA-secure encryption and CCA-secure KEM, in:
     Network Security, Springer, 2023, pp. 516–538, http://dx.doi.org/10.1007/978-               International Conference on Cryptology in Africa, Springer, 2018, pp. 282–305,
     3-031-33491-7_19.                                                                           http://dx.doi.org/10.1007/978-3-319-89339-6_16.
[13] K. Seyhan, S. Akleylek, Smaug kem to smaug-pake: a generic lattice-based pass-         [37] J. Bos, L. Ducas, E. Kiltz, T. Lepoint, V. Lyubashevsky, J.M. Schanck, P. Schwabe,
     word authenticated key exchange, Central European Conference on Cryptology                  G. Seiler, D. Stehlé, CRYSTALS-Kyber: a CCA-secure module-lattice-based KEM,
     CECC-2024 (2024) 38-41.                                                                     in: 2018 IEEE European Symposium on Security and Privacy (EuroS&P), IEEE,
[14] Z. Li, D. Wang, Achieving one-round password-based authenticated key exchange               2018, pp. 353–367, http://dx.doi.org/10.1109/EuroSP.2018.00032.
     over lattices, IEEE Trans. Serv. Comput. 15 (1) (2019) 308–321, http://dx.doi.         [38] E. Bresson, O. Chevassut, D. Pointcheval, Security proofs for an efficient
     org/10.1109/TSC.2019.2939836.                                                               password-based key exchange, in: Proceedings of the 10th ACM Conference on
[15] Z. Li, H. Zhu, G. Liao, M. Wang, P. Li, P. Gope, QT-PAKE: Secure messaging via              Computer and Communications Security, 2003, pp. 241–250, http://dx.doi.org/
     quantum-safe threshold PAKE, IEEE Trans. Consum. Electron. (2025).                          10.1145/948109.948142.
[16] J. Katz, V. Vaikuntanathan, Smooth projective hashing and password-based               [39] M. Bellare, D. Pointcheval, P. Rogaway, Authenticated key exchange secure
     authenticated key exchange from lattices, in: International Conference on the               against dictionary attacks, in: International Conference on the Theory and
     Theory and Application of Cryptology and Information Security, Springer, 2009,              Applications of Cryptographic Techniques, Springer, 2000, pp. 139–155, http:
     pp. 636–652, http://dx.doi.org/10.1007/978-3-642-10366-7_37.                                //dx.doi.org/10.1007/3-540-45539-6_11.


                                                                                       12
K. Seyhan et al.                                                                                                                  Computer Standards & Interfaces 97 (2026) 104118


[40] R. Canetti, S. Halevi, J. Katz, Y. Lindell, P. MacKenzie, Universally composable        [44] A. Jati, N. Gupta, A. Chattopadhyay, S.K. Sanadhya, A configurable crystals-kyber
     password-based key exchange, in: Annual International Conference on the Theory               hardware implementation with side-channel protection, ACM Trans. Embed.
     and Applications of Cryptographic Techniques, Springer, 2005, pp. 404–421,                   Comput. Syst. 23 (2) (2024) 1–25, http://dx.doi.org/10.1145/3587037.
     http://dx.doi.org/10.1007/11426639_24.                                                  [45] S. Aghapour, K. Ahmadi, M. Anastasova, M.M. Kermani, R. Azarderakhsh, PUF-
[41] A.F. Dursun, Smaug.PAKE for Mobile Devices, 2025, (Accessed: 24 October 2025)                kyber: Design of a PUF-based kyber architecture benchmarked on diverse ARM
     https://github.com/afDursun/lattice-based-pakes.                                             processors, IEEE Trans. Comput.-Aided Des. Integr. Circuits Syst. 43 (12) (2024)
[42] S. Aghapour, K. Ahmadi, M. Anastasova, R. Azarderakhsh, M. Mozaffari Kermani,                4453–4462, http://dx.doi.org/10.1109/TCAD.2024.3399669.
     PUF-dilithium: Design of a PUF-based dilithium architecture benchmarked on
     ARM processors, ACM Trans. Embed. Comput. Syst. 24 (2) (2025) 1–20, http:
     //dx.doi.org/10.1145/3715328.
[43] K. Ahmadi, S. Aghapour, M.M. Kermani, R. Azarderakhsh, Efficient error detec-
     tion cryptographic architectures benchmarked on FPGAs for montgomery ladder,
     IEEE Trans. Very Large Scale Integr. (VLSI) Syst. 32 (11) (2024) 2154–2158,
     http://dx.doi.org/10.1109/TVLSI.2024.3419700.




                                                                                        13
