Internet DRAFT - draft-lyu-privacy-enhanced-5g-aka

draft-lyu-privacy-enhanced-5g-aka







Network Working Group                                             X. Lyu
Internet-Draft                                                    X. Yao
Intended status: Standards Track                                   D. Ma
Expires: November 14, 2021                                        R. Hou
                                                                  J. Cao
                                                                  X. Mao
                                                       Xidian University
                                                            May 13, 2021


       PEAA: Privacy-Enhanced Access Authentication Scheme in 5G
                  draft-lyu-privacy-enhanced-5g-aka-00

Abstract

   Authentication and Key Agreement (AKA) protocol aims to enable mutual
   authentication between networks and a phone equipped with a Universal
   Subscriber Identity Module(USIM) card. 5G AKA introduces Subscription
   Permanent Identifier (SUPI) and Elliptic Curve Integrated Encryption
   Scheme (ECIES) to preserve user identity privacy.  However, when the
   authentication server becomes an unsecure entity due to malwares or
   malicious attacks, these existing schemes cannot effectively protect
   user identity privacy at the server-end.  For a small number of
   important users with high demand for identity privacy and security,
   comprehensive identity privacy protection is one of their important
   security requirements.  Based on a security assumption that the
   server is trustworthy within a limited time, this document proposes a
   privacy-enhanced access authentication scheme (PEAA) which is
   compatible with 5G AKA.  This scheme not only achieves user anonymity
   on wireless channels by dynamic temporary pseudonyms, but also
   ensures that the server authenticates users without their permanent
   identities.  In the context of hierarchical security requirements,
   PEAA scheme and the existing 5G AKA can respectively provide
   different user identity privacy protection services for users with
   different security requirements.

Status of This Memo

   This Internet-Draft is submitted in full conformance with the
   provisions of BCP 78 and BCP 79.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF).  Note that other groups may also distribute
   working documents as Internet-Drafts.  The list of current Internet-
   Drafts is at https://datatracker.ietf.org/drafts/current/.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any



Lyu, et al.             Expires November 14, 2021               [Page 1]

Internet-Draft                    PEAA                          May 2021


   time.  It is inappropriate to use Internet-Drafts as reference
   material or to cite them other than as "work in progress."

   This Internet-Draft will expire on November 14, 2021.

Copyright Notice

   Copyright (c) 2021 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (https://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

   This document may not be modified, and derivative works of it may not
   be created, except to format it for publication as an RFC or to
   translate it into languages other than English.

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   3
   2.  Acronyms and Symbols  . . . . . . . . . . . . . . . . . . . .   5
   3.  Proposed Scheme . . . . . . . . . . . . . . . . . . . . . . .   5
     3.1.  System Architecture . . . . . . . . . . . . . . . . . . .   6
     3.2.  Security Assumptions  . . . . . . . . . . . . . . . . . .   7
       3.2.1.  Adversary . . . . . . . . . . . . . . . . . . . . . .   7
       3.2.2.  Communication Entities  . . . . . . . . . . . . . . .   7
     3.3.  Security Requirements . . . . . . . . . . . . . . . . . .   8
     3.4.  Privacy-Enhanced Access Authentication Scheme . . . . . .   9
       3.4.1.  System Initialization . . . . . . . . . . . . . . . .  10
       3.4.2.  Offline Registration  . . . . . . . . . . . . . . . .  10
       3.4.3.  Real-time Authentication  . . . . . . . . . . . . . .  10
     3.5.  Apply PEAA scheme to Real Service Scenarios . . . . . . .  11
   4.  Security Analysis . . . . . . . . . . . . . . . . . . . . . .  11
     4.1.  Identity Authentication . . . . . . . . . . . . . . . . .  12
     4.2.  User Identity Privacy . . . . . . . . . . . . . . . . . .  12
       4.2.1.  User Identity Privacy on Wireless Channels  . . . . .  12
       4.2.2.  User Identity Privacy at the Server-end . . . . . . .  13
     4.3.  Confidentiality of system master key s  . . . . . . . . .  15
   5.  IANA Considerations . . . . . . . . . . . . . . . . . . . . .  16
   6.  Security Considerations . . . . . . . . . . . . . . . . . . .  16
   7.  References  . . . . . . . . . . . . . . . . . . . . . . . . .  17



Lyu, et al.             Expires November 14, 2021               [Page 2]

Internet-Draft                    PEAA                          May 2021


     7.1.  Normative References  . . . . . . . . . . . . . . . . . .  17
     7.2.  Informative References  . . . . . . . . . . . . . . . . .  17
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  18

1.  Introduction

   As one of the most widely used Internet access interfaces, mobile
   communication networks have shown great success in many aspects, such
   as communicating with others, paying with mobile phones and
   supporting the Internet of Things (IoT).  Users are connected to
   mobile networks by devices equipped with their own USIM cards, which
   is achieved by variants of AKA protocol in 3G and 4G mobile networks.
   The 5th Generation mobile networks (5G) and telecommunication
   standard are currently under development, and are nearly finalized.
   The identity authentication in 5G system is based on new versions of
   AKA protocols, which are included in [TS33501], notably the new 5G
   AKA protocol. 5G AKA protocol, which involves User Equipment(UE),
   Serving Network(SN), and Home Network(HN), is an evolution of the AKA
   variants used in 3G and 4G, and it aims to implement mutual
   authentication between UE and HN and complete session key agreement
   between UE and SN.  UE communicates with SN through insecure wireless
   channels, while SN communicates with HN through secure wired
   channels.  The overall architecture of 5G AKA is as follows:

     +---------+                 +---------+                +---------+
     |   User  |insecure wireless| Serving |  secure wired  |  Home   |
     |Equipment|<--------------->| Network |<-------------->| Network |
     |   (UE)  |   channels      |  (SN)   |    channels    |   (HN)  |
     +---------+                 +---------+                +---------+

                 Figure 1: Overall Architecture of 5G AKA

   Due to the openness of wireless channels, leaking user identity is
   possible during the authentication process [Ahlawat18].  Once an
   adversary obtains a UE's unique identity------International Mobile
   Subscriber Identity (IMSI), he/she can get the UE's moving trajectory
   and keep track of the UE or even forge the UE's request.  In recent
   years, how to effectively protect user identity privacy has become an
   important research hotspot in the field of mobile communication
   authentication [Ferrag18] [Khan18], and 3GPP is gradually improving
   user identity privacy-preserving mechanism. 3G system introduces
   Temporary Mobile Subscriber Identity (TMSI) to protect UE's real
   identity IMSI.  However, a UE's TMSI is only valid in a given Visitor
   Location Register (VLR).  When the current TMSI expires or the UE
   needs to access a new VLR because of UE's updated location, the UE
   needs to send its IMSI to the current VLR to get a new TMSI.  In this
   process, there is a security risk of identity leakage.  Based on the
   idea of TMSI in 3G system, 4G system adopts Globally Unique Temporary



Lyu, et al.             Expires November 14, 2021               [Page 3]

Internet-Draft                    PEAA                          May 2021


   Identifier (GUTI).  However, UE needs to send its IMSI for
   authentication and obtain its GUTI when UE accesses networks for the
   first time.  In this process, UE's IMSI may be leaked.

   In order to meet the requirements of protecting user identity
   privacy, 5G system replaces IMSI with Subscription Permanent
   Identifier (SUPI) [TS33501], that is, SUPI is the permanent real user
   identity in 5G system.  Considering the privacy of SUPI, 5G system
   utilizes ECIES, an asymmetric cryptographic algorithm, to preserve
   user identity privacy.  In the ECIES-based scheme, UE and HN use
   public/private key pairs and other parameters to complete key
   agreement, and then generate a symmetric encryption key.  UE uses the
   key to encrypt SUPI to obtain Subscription Concealed Identifier
   (SUCI) and then sends SUCI to HN.  HN uses the key to decrypt SUCI to
   get SUCI.  So 5G AKA can protect user identity privacy on wireless
   channels.  But this scheme still needs improvement: Only when HN
   acquires a UE's real identity SUPI, can it verify whether the UE is
   legal or not.  However, there may be some security risks when HN
   knows UE's real identity, because HN is not always in a secure and
   reliable state [FRoE17], for example, it is possible that HN is
   implanted with malicious malware, which may leak out UE's SUPI.  In
   this case, this scheme cannot protect user identity privacy at the
   server-end.  In addition, [Gharsallah19] pointed out that the
   introduction of asymmetric cryptographic algorithm ECIES requires
   Public Key Infrastructure (PKI), which may increase the computational
   overhead of users and authentication servers.

   In order to solve these problems, this document proposes a privacy-
   enhanced access authentication scheme (PEAA), the key points are as
   follows:

   o  PEAA scheme can ensure that HN can authenticate a UE without
      knowing the UE's unique permanent identity SUPI.  It means that
      the scheme can preserve user identity privacy at the server-end.

   o  User identity privacy at the server-end can still be protected
      under typical attacks (e.g., replay attacks), even if the server-
      end is invaded by a malicious adversary during the process of
      authentication.

   o  PEAA scheme is compatible with 5G AKA.  In the context of
      hierarchical security requirements, PEAA scheme and the existing
      5G AKA can respectively provide different user identity privacy
      protection services for users with different security
      requirements.  For users with general requirements for identity
      privacy-preserving, the existing 5G authentication scheme can be
      used to focus on protecting user identity privacy on wireless
      channels; for a few important users with high-level security



Lyu, et al.             Expires November 14, 2021               [Page 4]

Internet-Draft                    PEAA                          May 2021


      requirements, PEAA scheme can effectively and comprehensively
      protect user identity privacy.

   o  PEAA scheme can complete user identity authentication without
      additional authentication processes or authentication entities
      (e.g., PKI).

   In this document, Section 2 introduces the involved acronyms and the
   meaning of some symbols.  Section 3 introduces PEAA scheme in detail,
   including system architecture, security assumptions, security
   requirements, detailed authentication process and so on.  Section 4
   analyzes the security of PEAA scheme.

2.  Acronyms and Symbols

   This document uses the following acronyms:

     3GPP   3rd Generation Partnership Project
     5G     5th Generation mobile network
     AKA    Authentication and Key Agreement
     DCDH   Divisible Computation Diffie-Hellman Assumption
     ECIES  Elliptic Curve Integrated Encryption Scheme
     GUTI   Globally Unique Temporary Identity
     HN     Home Network
     HSS    Home Subscriber Server
     IAP    Identity Authentication Pair
     IMSI   International Mobile Subscriber Identity
     IoT    Internet of Things
     PEAA   privacy-enhanced access authentication scheme
     PKI    Public Key Infrastructure
     SN     Serving Network
     SUCI   Subscription Concealed Identifier
     SUPI   Subscription Permanent Identifier
     TMSI   Temporary Mobile Subscriber Identity
     UE     User Equipment
     USIM   Universal Subscriber Identity Module
     VLR    Visitor Location Register

   In this document, "<-" stands for "belong to".  For example, "a <-
   A", which means that element a belongs to set A.  The meaning of "!="
   is "not equal to". "^" stands for exponential operation, and "Zp*"
   implies a multiplicative group {1, 2, 3, ..., p-1}.

3.  Proposed Scheme

   In this section, PEAA scheme is introduced in detail.  This section
   first emphasizes the system architecture and the security assumptions
   of this scheme, then introduces the detailed authentication process



Lyu, et al.             Expires November 14, 2021               [Page 5]

Internet-Draft                    PEAA                          May 2021


   of PEAA scheme, and finally considers some practical application
   scenarios.

3.1.  System Architecture

   To ensure that the scheme is compatible with the existing 5G AKA, the
   communication entities included in PEAA scheme are all defined
   according to 5G standard.  The entities included in this scheme are
   User Equipment (UE), Serving Network (SN), and Home Subscription
   Server (HSS).

   o  User Equipment (UE): A device that allows a user to access network
      services.  Each UE has its own SUPI which is stored in UE's USIM
      card.  In this document, UE has the same meaning as user for
      convenience.

   o  Service Network (SN): A transit site that provides UE with network
      services.  According to 3GPP standard, UE communicates with SN
      through wireless channels, while SN communicates with HN through
      secure wired channels.  Considering that SN plays a role of
      transmitting authentication information in the authentication
      process, SN does not affect encryption operations and decryption
      operations in the authentication process.  So this scheme only
      considers UE and HSS in the subsequent discussion.

   o  Home Subscriber Server (HSS): The core server that stores users'
      data (e.g., user identity) and authentication information.  In
      this scheme, HSS authenticates user identity, and it is equivalent
      to the server-end or the above-mentioned HN.

   This scheme mainly includes three stages:

   o  System Initialization: HSS selects system parameters required for
      identity authentication;

   o  Offline Registration: UE performs identity registration at the
      HSS-end and obtains required information for identity
      authentication from HSS;

   o  Real-time Authentication: When UE needs the services provided by
      SN, SN initiates user identity authentication, and HSS verifies
      the legitimacy of the user identity authentication request.

   In the first two stages, USIM card has not been allocated to UE;
   after UE obtains its USIM card from the mobile network service
   provider, the subsequent identity authentication process is
   performed.




Lyu, et al.             Expires November 14, 2021               [Page 6]

Internet-Draft                    PEAA                          May 2021


3.2.  Security Assumptions

   This section will describe security assumptions from the following
   two aspects: adversary and communication entities.

3.2.1.  Adversary

   A privacy-preserving scheme is designed to reduce the sensitive
   information acquired by adversaries while ensuring the quality of
   service to users.  Based on the sensitive information that is
   available to an adversary, we defined two types of adversaries in
   this document: passive adversary and active adversary.

   o  Passive Adversary: A passive adversary can only wiretap insecure
      channels.  Although the adversary can get encrypted information by
      eavesdropping on insecure channels, he/she cannot decrypt it.
      Based on the eavesdropped information, he/she tries to infer some
      sensitive information of a particular user, such as user identity,
      sensitive locations.

   o  Active Adversary: An active adversary can actively invade SN or
      HSS, and may obtain real identity of a target user through
      decryption if he/she has enough decrypted information.

   In this document, both passive adversary and active adversary aim to
   obtain users' real identities.  Based on this security assumption,
   adversaries will only launch attacks with an ultimate goal of
   obtaining users' real identities.  This scheme does not consider
   other attacks whose purpose is not to obtain users' real identities,
   for example, an adversary blocks a user's authentication request to
   disrupt the user's normal authentication process.

3.2.2.  Communication Entities

   PEAA scheme relies on a security assumption that HSS is trustworthy
   within a limited time, so this section will clarify the security
   status of each communication entity in each authentication stage so
   as to facilitate subsequent security analysis.  For the communication
   entities involved in this scheme, the security assumptions are as
   follows:

   o  HSS is secure and trustworthy during the processes of System
      Initialization and Offline Registration.  Since System
      Initialization and Offline Registration are completed before UE
      securely obtains its USIM card, HSS has unilateral control in
      these two stages.  So this scheme assumes that HSS is trustworthy
      during System Initialization and Offline Registration;




Lyu, et al.             Expires November 14, 2021               [Page 7]

Internet-Draft                    PEAA                          May 2021


   o  UE is secure during the processes of System Initialization and
      Offline Registration.  USIM card has not been allocated to UE in
      these two stages.  At this time, UE is in a state of securely
      receiving the necessary authentication information from HSS.  So
      this scheme assumes that UE is secure during these two stages.

   o  HSS is insecure during the process of Real-time Authentication.
      HSS is vulnerable to malicious attacks because HSS's identity,
      location, and other information are public.  If an adversary
      invades HSS and obtains key authentication information, HSS will
      be in an insecure state and may reveal users' identities during
      the subsequent authentication process.  PEAA scheme focuses on
      user identity privacy-preserving at the server-end, so this scheme
      assumes that HSS is in an insecure state during the process of
      Real-time Authentication;

   o  UE is secure during the process of Real-time Authentication.  If
      UE is in an insecure state during the process of Real-time
      Authentication, an adversary may invade UE and obtain its real
      identity.  At this time, the focus of protecting user identity
      privacy is to enhance the security of preset authentication
      information in UE, not the security of identity authentication
      scheme.  It is inconsistent with the focus of PEAA scheme, so this
      scheme assumes that UE is in a secure state during the process of
      Real-time Authentication.

3.3.  Security Requirements

   A privacy-enhanced access authentication scheme should meet the
   following security requirements:

   o  Identity authentication: It should be guaranteed that HSS can
      authenticate the legitimacy of UE's identity and authorize SN to
      provide legitimate UE with corresponding network services;

   o  Variable temporary identities: It should be guaranteed that UE can
      use different dynamic temporary identities in different
      authentications.

   o  Untraceability: It should be guaranteed that different temporary
      identities used by a UE in different authentication processes
      cannot be correlated, and the temporary identity and the UE's real
      identity cannot be correlated either.

   o  User identity privacy-preserving on wireless channels: It should
      be guaranteed that an adversary cannot obtain UE's SUPI by
      eavesdropped authentication message on wireless channels.




Lyu, et al.             Expires November 14, 2021               [Page 8]

Internet-Draft                    PEAA                          May 2021


   o  User identity privacy-preserving at the server-end: It should be
      guaranteed that HSS can verify the legitimacy of UE's identity
      without knowing UE's SUPI.

3.4.  Privacy-Enhanced Access Authentication Scheme

   Compared with the existing authentication schemes, PEAA scheme not
   only pays attention to the security of user identity on wireless
   channels, but also preserves user anonymity at the server-end.  PEAA
   scheme comprises three stages: System Initialization, Offline
   Registration, Real-time Authentication.  The specific procedures of
   this scheme is shown in Figure 2.

     UE                                                           HSS
   (K,SUPI)                                                   (K,SUPI,s)
     |                        1. System Initialization              |
     |             _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _  |
     |            |HSS selects generator g and system master key s| |
     |            |_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _| |
     |                                                              |
     |                      2. Offline Registration                 |
     |  _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _   |
     | |--------------------      SUPI, K    -------------------->| |
     | |<-------------------  g, SUPI^s, K^s ---------------------| |
     | |_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ | |
     |                                                              |
     |                     3. Realtime Authentication               |
     |  _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |
     | |1) UE selects secret parameter u and temporary mask r     | |
     | |                                                          | |
     | |2) UE caculates AUTH_req and then send it to HSS.         | |
     | |...................................................       | |
     | |: AUTH_req = {SUPI^r, g^r, g^u, (Pi1,Pi2), Ts, h} :       | |
     | |: h = H(SUPI^(s*r)||(Pi1,Pi2)||Ts)                :------>| |
     | |: Pi1 = K^(u*s) + SUPI^(s*r),Pi2 = K^(u*(SUPI^r)) :       | |
     | |...................................................       | |
     | |                                                          | |
     | |           3) HSS performs integrity check:               | |
     | |           P = (SUPI^r)^s;h' = H(P||(Pi1,Pi2)||Ts);h' = h | |
     | |                                                          | |
     | |           4) HSS verifies the identity validity of UE:   | |
     | |           (Pi1-SUPI^(s*r))^(SUPI^r) = Pi2^s              | |
     | |_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ | |
     |                                                              |
     | <-------HSS verifies the identity validity of the UE---------|

                  Figure 2: The Procedures of PEAA scheme




Lyu, et al.             Expires November 14, 2021               [Page 9]

Internet-Draft                    PEAA                          May 2021


3.4.1.  System Initialization

   Given a cyclic group G of prime order p.  HSS selects a generator g
   <- G of group G, and chooses s <- Zp* as system master key.  Then HSS
   makes the generator g public and keeps the system master key s
   secret.  Note that s != 1 mod p and s != 0 mod p for security
   reasons.

3.4.2.  Offline Registration

   Referring to Section 3.2.2, HSS is trustworthy during the process of
   Offline Registration, that is, UE can register at the HSS-end
   securely.

   o  UE --> HSS: UE sends SUPI mod p and the corresponding pre-shared
      symmetric key {K mod p} to HSS;

   o  HSS --> UE: HSS calculates SUPI^s mod p and K^s mod p, and sends
      the results to UE.  Since all subsequent authentication operations
      are performed on the cyclic group G, the symbol "mod p" is omitted
      in order to simplify expression.

3.4.3.  Real-time Authentication

   When UE obtains the identity authentication information and needs to
   use network services, SN forwards UE's authentication request, then
   HSS verifies the request and instructs SN whether to provide network
   services for UE.  The specific authentication process is as follows:

   o  UE selects parameters: UE selects a fixed secret parameter u <-
      Zp* (u != 1).  Meanwhile, UE selects r <- Zp* (r != 1) as a
      temporary mask.

   o  UE --> HSS: UE calculates SUPI^r as a temporary pseudonym and
      obtains AUTH_req according to Equation 1.  Then UE sends AUTH_req
      to HSS.  This document defines (Pi1, Pi2) as Identity
      Authentication Pair (IAP).

       +--------------------------------------------------------+
       | AUTH_req = {SUPI^r, g^r, g^u, (Pi1,Pi2), Ts, h}        |
       |                                                        |
       | h = H(SUPI^(s*r)||(Pi1,Pi2)||Ts)                       |
       |                                             Equation 1 |
       | Pi1 = K^(u*s) + SUPI^(s*r)                             |
       |                                                        |
       | Pi2 = K^(u*(SUPI^r))                                   |
       +--------------------------------------------------------+




Lyu, et al.             Expires November 14, 2021              [Page 10]

Internet-Draft                    PEAA                          May 2021


   o  HSS performs integrity check: After receiving AUTH_req = {SUPI^r,
      g^r, g^u, (Pi1,Pi2), Ts, h} from UE, HSS calculates P=(SUPI^r)^s
      according to system master key s and the received SUPI^r, and then
      calculates h' = H(P||(Pi1,Pi2)||Ts).  If h' != h, the received
      AUTH_req isn't integrate, so HSS ignores the authentication
      request and sends a failure signal to UE who sends AUTH_req; if h'
      = h, the received AUTH_req is integrate, and then HSS continues to
      perform the subsequent fourth step.

   o  HSS verifies the identity validity of UE: HSS calculates
      (Pi1-SUPI^(s*r))^(SUPI^r) and judges whether it is equal to Pi2^s.
      If (Pi1-SUPI^(s*r))^(SUPI^r) != Pi2^s, it means that AUTH_req does
      not contain the legal user identity, and the authentication fails;
      if (Pi1-SUPI^(s*r))^(SUPI^r) = Pi2^s, it means that the temporary
      pseudonym SUPI^r contained in the authentication tuple represents
      the real identity of UE who sends AUTH_req, and HSS verifies the
      identity validity of the UE.  SN is authorized to supply
      corresponding services to the legitimate UE who has the temporary
      pseudonym SUPI^r.

3.5.  Apply PEAA scheme to Real Service Scenarios

   In a lot of real application scenarios, HSS needs to know stable
   identities of users so that it can provide services for UE, such as
   location-based services, call forwarding services, billing service.
   Especially for the billing service, HSS needs to link a fixed billing
   account with a UE, and determines whether to provide the UE with
   communication services based on the balance of UE's account.  Thus
   there is a conflict, that is, UE wants to preserve user identity
   privacy at the HSS-end, but HSS needs to obtain UE's stable identity
   in order to provide corresponding services.  To resolve this
   conflict, this scheme maps UE's SUPI to an anonymous fixed billing
   account.  Shown in Section 3.4.3, Pi1 = K^(u*s) + SUPI^(s*r).  When
   HSS receives Pi1 and SUPI^r, HSS can get UE's stable anonymous
   identity K^(u*s).  Noted that getting fixed secret parameter u in the
   case of knowing K^(u*s) and system master key s is a Discrete
   Logarithm(DL) Problem.  HSS can assign a stable billing account to
   the anonymous identity K^(u*s).  By mapping UE's SUPI to K^(u*s),
   this scheme not only protects user identity privacy at the HSS-end,
   but also enables HSS to provide normal billing service without
   revealing UE's SUPI.

4.  Security Analysis

   This section will analyze the security of PEAA scheme from three
   aspects: Identity Authentication, User Identity Privacy,
   Confidentiality of system master key s.




Lyu, et al.             Expires November 14, 2021              [Page 11]

Internet-Draft                    PEAA                          May 2021


4.1.  Identity Authentication

   By judging whether h' = H(P||(Pi1, Pi2)||Ts) is equal to h and (Pi1-
   SUPI^(s*r))^(SUPI^r) is equal to Pi2^s, HSS can check whether UE has
   SUPI^s and the corresponding K^s.  According to the security
   assumption in Section 3.2.2, UE can keep its SUPI and the
   corresponding pre-shared symmetric key K secret, that is, USIM card
   has no security risks, and HSS can securely allocate SUPI^s and K^s
   to UE.  If a UE can provide an IAP containing both SUPI^s and K^s,
   HSS can authenticate the UE's legitimacy, and then SN is authorized
   to provide network services for the UE.  Otherwise, the
   authentication fails.

4.2.  User Identity Privacy

   This section will analyze whether user identity privacy can be
   preserved from the following two aspects: one is user identity
   privacy on wireless channels, the other is user identity privacy at
   the server-end.  Correspondingly, this section defines two types of
   adversaries: Adversary AI and Adversary AII.

4.2.1.  User Identity Privacy on Wireless Channels

   Adversary AI is a passive adversary, who can eavesdrop on wireless
   channels with security risks and obtain a series of information
   exchanged by a target user with HSS during their authentication
   process.  Assume that AI is also a legitimate user in the
   communication system, and performs normal identity authentication
   like other legitimate users.  Based on the eavesdropped
   authentication information about the target user and the information
   exchanged by AI with HSS during their authentication process, AI
   tries to infer SUPI of the target user.  For AI, the process to
   obtain these information is as follows:

   o  Adversary AI performs normal identity authentication:

      *  System Initialization: HSS selects generator g <- G of group G,
         and chooses s as system master key.  Then HSS sends the
         generator g to Adversary AI and keeps the system master key s
         secret.

      *  Offline Registration: Adversary AI sends his/her real identity
         SUPI_ad and the corresponding pre-shared symmetric key K_ad to
         HSS, HSS calculates SUPI_ad^s and K_ad^s, and sends the results
         to Adversary AI.

      *  Real-time Authentication: Adversary AI sends AUTH_req_ad to
         HSS, and HSS returns the verification result of AUTH_req_ad;



Lyu, et al.             Expires November 14, 2021              [Page 12]

Internet-Draft                    PEAA                          May 2021


   o  Adversary AI eavesdrops on wireless channels: Adversary AI can
      obtain the information exchanged between his/her target user and
      HSS during the process of Real-time Authentication by
      eavesdropping on wireless channels.  The target user, whose real
      identity is SUPI, will send AUTH_req_user to HSS during the
      process of real-time authentication, that is, Adversary AI can
      obtain AUTH_req_user that is sent by the target user on wireless
      channels.

   In summary, Adversary AI has already obtained {g, SUPI_ad, SUPI_ad^s,
   K_ad, K_ad^s, AUTH_req_ad, AUTH_req_user}, and then AI tries to
   obtain the target user's real identity SUPI.

   Theorem 4.1.  If adversary AI can extract real identity SUPI of the
   target user from {g, AUTH_req_user} with the probability P in time t,
   HSS can solve Divisible Computation Diffie-Hellman (DCDH) problem
   with the probability P' = P in time t' = t.

   PROOF.  Referring to Section 3.4, g is the generator of G.  Given (g,
   g^x, g^y), where x, y <- Zq*, the process of HSS calculating g^(x/y)
   is as follows:

   o  HSS runs AI with {SUPI^r = g^x, g^r = g^y} to obtain SUPI of the
      target user, and the probability of success is P and time of the
      process is t;

   o  Referring to Equation 1, HSS calculates like this: SUPI^r = g^x
      ==> SUPI = g^(1/r*x) ==> SUPI = g^(x/y), the probability of
      success is 1, and the time spent is ignored;

   o  Finally, HSS uses (g, g^x, g^y) to obtain g^(x/y) with the
      probability P' = P*1 = P, and the total time is t'= t.

   Conclusion: For Adversary AI on wireless channels, obtaining real
   identity SUPI of the target user is equivalent to the process that
   HSS solve the DCDH problem.

4.2.2.  User Identity Privacy at the Server-end

   Adversary AII is an active adversary.  He/she can actively invade HSS
   and makes HSS be in an insecure state during the process of Real-time
   Authentication.  At this time, Adversary AII can obtain the
   information stored by HSS, such as HSS's system master key s,
   authentication information sent by Adversary AII's target user to HSS
   and so on.  Assume that AII is also a legitimate user in the
   communication system, and performs normal identity authentication
   like other legitimate users.  Based on the information obtained by
   invading HSS and the information exchanged by AII with HSS during



Lyu, et al.             Expires November 14, 2021              [Page 13]

Internet-Draft                    PEAA                          May 2021


   their authentication process, AII tries to infer SUPI of the target
   user.  For AII, the process to obtain these information is as
   follows:

   o  Adversary AII performs normal identity authentication: This
      process is the same as the process described in Section 4.2.1, so
      this document won't repeat it here.

   o  Adversary AII invades HSS: After invading HSS, Adversary AII can
      obtain the authentication information AUTH_req_user sent by the
      target user to HSS and the system master key s of HSS.

   In summary, Adversary AII has already obtained {g, s, AUTH_req_user},
   and then AII tries to obtain the target user's real identity SUPI.
   It should be noted that legitimate users will send their SUPI and the
   corresponding pre-shared symmetric key K to HSS during the process of
   Offline Registration, which means that HSS stores all legitimate
   users' SUPI, corresponding pre-shared symmetric key K, and
   corresponding relationship between SUPI and K, which are called
   "SUPI-K Related Information" for convenience.  Considering that HSS
   is invaded, SUPI-K Related Information may also be obtained by
   Adversary AII.  Based on this assumption, if AII can obtain pre-
   shared symmetric key K of the target user on the premise of acquiring
   {g, s, AUTH_req_user}, AII can query SUPI-K related information to
   get SUPI of the target user.  This document will analyze the
   situation where Adversary AII makes use of {g, s, AUTH_req_user} to
   obtain pre-shared symmetric key K of the target user.

   Theorem 4.2.  If Adversary AII can extract pre-shared symmetric key K
   from {g, s, AUTH_req_user} with the probability P in time t, HSS can
   solve DCDH problem with the probability P' = P in time t' = t.

   PROOF.  Referring to Section 3.4, g is the generator of G.  Given (g,
   g^x, g^y), where x, y <- Zq*, the process of HSS calculating g^(x/y)
   is as follows:

   o  HSS runs AII with {s, K^(u*s) = g^x, g^u = g^y} to obtain pre-
      shared symmetric key K of the target user, and the probability of
      success is P and time of the process is t;

   o  Referring to Equation 1, HSS calculates like this: K^(u*s) = g^x
      ==> K^s = g^(1/u*x) ==> K^s = g^(x/y) ==> K = g^(x/(s*y)), the
      probability of success is 1, and the time spent is ignored;

   o  Finally, HSS uses (g, g^x, g^y) to obtain g^(x/y) with the
      probability P' = P*1 = P, and the total time is t'= t.





Lyu, et al.             Expires November 14, 2021              [Page 14]

Internet-Draft                    PEAA                          May 2021


   Conclusion: For the adversary AII who invades HSS, obtaining the
   target user's pre-shared symmetric key K is equivalent to the process
   that HSS solve the DCDH problem.

4.3.  Confidentiality of system master key s

   Referring to the security assumption about HSS in Section 3.2.2, HSS
   is in an insecure state due to being invaded by an adversary or being
   implanted with malwares in the stage of Real-time Authentication.  At
   this point, the adversary can obtain the system master key s of HSS
   and has the ability to decrypt.  He/she also has the ability to forge
   a series of AUTH_req and send them to HSS to launch Denial of
   Service(DoS) attacks.  However, the security assumption in
   Section 3.2.1 clearly points out that adversaries will only launch
   attacks with an ultimate goal of obtaining users' real identities,
   and this scheme does not consider other attacks whose purpose is not
   to obtain users' real identities.  So this document only considers
   whether a legitimate UE can obtain the system master key s of HSS by
   decrypting the existing authentication message without additional
   information.  In other words, a legitimate UE holds {g, SUPI, K,
   SUPI^s, K^s}and wants to extract HSS's system master key s.  For
   convenience, this document only considers the situation where the
   legitimate UE makes use of {g, SUPI, SUPI^s} to extract HSS's system
   master key s.

   Theorem 4.3.  If a legitimate UE can extract HSS's system master key
   s from {g, SUPI, SUPI^s} with the probability P in time t, HSS can
   solve DCDH problem and DL problem with the probability P' = P in time
   t' = t.

   PROOF.  Referring to Section 3.4, g is the generator of G.  Given (g,
   g^x, g^y), where x, y <- Zq*, the process of HSS calculating g^(x/y)
   is as follows:

   o  HSS runs UE with {g, SUPI = g^x, SUPI^s = g^y} to obtain the
      system master key s of HSS, and the probability of success is P
      and time of the process is t;

   o  Referring to Equation 1, HSS calculates like this: g^y = g^(s*x)
      ==> g^s = g^(y/x).  First, UE calculates g^(y/x) according to (g,
      g^x, g^y), and then calculates the system master key s according
      to (g, g^s);

   o  Finally, HSS uses (g, g^x, g^y) to obtain system master key s with
      the probability P' = P, and the total time is t'= t.






Lyu, et al.             Expires November 14, 2021              [Page 15]

Internet-Draft                    PEAA                          May 2021


   Conclusion: For a legitimate UE with existing certification
   information, obtaining HSS's system master key s is equivalent to the
   process that HSS solve the DCDH problem and DL problem.

5.  IANA Considerations

   There is no IANA consideration needed.

6.  Security Considerations

   In order to resist replay attacks, this scheme uses a time stamp Ts,
   which describes the life cycle of the current authentication message,
   and defines the integrity check value h = H(SUPI^(s*r)||(Pi1,
   Pi2)||Ts), where SUPIs is part of UE's secret storage.  Therefore, on
   the premise of only obtaining SUPI^r, it is difficult for an
   adversary to forge legal AUTH_req without changing integrity check
   value h.  But according to the security assumption in Section 3.2.2,
   HSS may be in an insecure state, and the adversary may obtain HSS's
   system master key s.  In this case, replay attacks cannot be avoided.
   However, UE's SUPI won't be leaked under replay attacks, and user
   identity privacy can be preserved without affecting the normal
   authentication of target users.

   In addition to replay attacks, an adversary may launch masquerading
   attacks, that is, he/she tries to extract IAP = (Pi1, Pi2) from the
   old legitimate authentication request of the target user.  Referring
   to Equation 1, the adversary cannot obtain the legal IAP without
   HSS's system master key s.  As with replay attacks, when the
   adversary obtains system master key s of HSS, the target user cannot
   avoid impersonation attacks, but user identity privacy still can be
   preserved.

   In summary, when HSS's system master key s is kept secret, PEAA
   scheme can resist replay attacks and masquerading attacks.  PEAA
   scheme can protect user identity privacy at the HSS-end, but cannot
   resist these two attacks when HSS becomes insecure due to malwares or
   malicious attacks and the system master key s is leaked.  As a
   consequence, the key to avoiding replay attacks and impersonation
   attacks is to protect HSS's system master key s.  In this scheme, HSS
   regularly updates its system master key s.  UE needs to perform
   Offline Registration regularly, and HSS is responsible for updating
   authentication-related information.

   Finally, an adversary may send a large number of authentication
   requests to consume computing resources and bandwidth resources of
   HSS as much as possible, thereby preventing other legitimate users
   from accessing.  In this scheme, after receiving authentication
   request AUTH_req from UE, HSS performs exponentiation and hashing



Lyu, et al.             Expires November 14, 2021              [Page 16]

Internet-Draft                    PEAA                          May 2021


   operations to check the integrity of AUTH_req.  Compared to
   asymmetric decryption, exponentiation and hashing operations are
   lightweight.  After confirming the integrity of AUTHreq, HSS will
   proceed with subsequent heavyweight certification operations.
   Although the Denial of Service(DoS) attacks cannot be completely
   resisted, the impact of DoS attacks can be reduced with the
   preliminary integrity check.

7.  References

7.1.  Normative References

   [TS33501]  3rd Generation Partnership Project, "Security architecture
              and procedures for 5G system", Release 16, TS 33.501,
              September 2019.

7.2.  Informative References

   [Ahlawat18]
              Ahlawat, A. and S. Kumar, "Investigating Various Possible
              Attacks and Vulnerabilties in LTE. International Journal
              of Computerences & Engineering, Vol. 6 (3).", 2018.

   [Ferrag18]
              Ferrag, M., Maglaras, L., Argyriou, A., Kosmanos, D., and
              H. Janicke, "Security for 4G and 5G cellular networks: A
              survey of existing authentication and privacy-preserving
              schemes. Journal of Network and Computer Applications",
              2018.

   [FRoE17]   3rd Generation Partnership Project, "First response on
              ECIES for concealing IMSI or SUPI, 3GPP TSG SA WG3
              (Security) Meeting", 2017.

   [Gharsallah19]
              Gharsallah, I., Smaoui, S., and F. Zarai, "A secure
              efficient and lightweight authentication protocol for 5G
              cellular networks: SEL-AKA. 2019 15th International
              Wireless Communications & Mobile Computing Conference
              (IWCMC)", June 2019.

   [Khan18]   Khan, M., Niemi, V., and P. Ginzboorg, "IMSI-based Routing
              and Identity Privacy in 5G", 2018.








Lyu, et al.             Expires November 14, 2021              [Page 17]

Internet-Draft                    PEAA                          May 2021


Authors' Addresses

   Xixiang Lyu
   Xidian University
   No.266 Xifeng Road
   Shanxi, Xi'an  710126
   China

   Email: xxlv@mail.xidian.edu.cn


   Xinyue Yao
   Xidian University
   No.266 Xifeng Road
   Shanxi, Xi'an  710126
   China

   Email: xyyaoo@stu.xidian.edu.cn


   Dong Ma
   Xidian University
   No.266 Xifeng Road
   Shanxi, Xi'an  710126
   China

   Email: 1282844751@qq.com


   Ronghui Hou
   Xidian University
   No.266 Xifeng Road
   Shanxi, Xi'an  710126
   China

   Email: rhhou@xidian.edu.cn


   Jin Cao
   Xidian University
   No.266 Xifeng Road
   Shanxi, Xi'an  710126
   China

   Email: jcao@xidian.edu.cn






Lyu, et al.             Expires November 14, 2021              [Page 18]

Internet-Draft                    PEAA                          May 2021


   Xinhong Mao
   Xidian University
   No.266 Xifeng Road
   Shanxi, Xi'an  710126
   China

   Email: 16145600@qq.com












































Lyu, et al.             Expires November 14, 2021              [Page 19]