hash
stringlengths
32
32
doc_id
stringlengths
5
12
section
stringlengths
5
1.47k
content
stringlengths
0
6.67M
837a5602c57c068290b1254827b20390
33.700-23
5.4.3 Potential security requirements
Editor’s Note: Potential security requirements are FFS. 5.X Key issue #X: <Title> 5.X.1 Key issue details 5.X.2 Threats 5.X.3 Potential security requirements
837a5602c57c068290b1254827b20390
33.700-23
6 Proposed solutions
837a5602c57c068290b1254827b20390
33.700-23
6.0 Mapping of solutions to key issues
Table 6.0-1: Mapping of solutions to key issues Solutions KI#1 KI#2 KI#3 Solution #1 X Solution #2 X Solution #3 X Solution #4 X Solution #5 X Solution #6 X Solution #7 X Solution #8 X Solution #9 X Solution #10 X Solution #11 X
837a5602c57c068290b1254827b20390
33.700-23
6.1 Solution #1: Addressing security aspects of "UE-deployed API invoker accessing other UEs’ resources of a group" procedure
837a5602c57c068290b1254827b20390
33.700-23
6.1.1 Introduction
This solution addresses key issue #1 (Group Authorization for UE-deployed API invoker accessing other UEs' resources of a group) by taking the procedure specified in clause 8.34 of TS 23.222 [2] as the baseline. As stated in the specified procedure, how to obtain authorization data from the GRO is out of scope, which means that the specification has the assumption that the authorization data is available at the CCF. This solution is also based on that assumption.
837a5602c57c068290b1254827b20390
33.700-23
6.1.2 Solution details
Security related addition to the procedure specified in clause 8.34 of TS 23.222 [2] is shown below. - In step 2 of the procedure in clause 8.34.3 of TS 33.222 [2], the CCF also obtains the GPSI of UE2 (API Invoker) in an authenticated way and uses that authenticated UE2 GPSI information in step 3. This solution does not describe how the CCF obtains the GPSI of UE2 in an authenticated way and proposes to leave it to implementation. Editor’s Note: Whether current methods are enough for group authorization is enough is FFS.
837a5602c57c068290b1254827b20390
33.700-23
6.1.3 Evaluation
Editor's Note: Evaluation is FFS.
837a5602c57c068290b1254827b20390
33.700-23
6.2 Solution #2: Security aspect of group authorization
837a5602c57c068290b1254827b20390
33.700-23
6.2.1 Introduction
This solution addresses KI#1: Group Authorization for UE-deployed API invoker accessing other UEs' resources of a group. The existing API invoker authorization mechanism for RNAA is enhanced to support group authorization.
837a5602c57c068290b1254827b20390
33.700-23
6.2.2 Solution details
This solution reuses the procedure of UE-deployed API invoker accessing other UEs’ resources of a group defined in clause 8.34.3 of TS 23.222 [2]. Figure 6.2.2-1: Group Authorization mechanism for UE-deployed API invoker accessing other UEs' resources of a group 1. With reference to step 1 in clause 8.34.3 of TS 23.222 [2], the request is formatted as an OAuth 2.0 access token request. 2. With reference to step 2 in clause 8.34.3 of TS 23.222 [2], CCF performs authentication of the API invoker by verifying the API invoker’s credentials. 3. With reference to step 3 in clause 8.34.3 of TS 23.222 [2], CCF additionally checks whether the UE whose resources are to be accessed belongs to the group. 4. Same as step 4 in clause8.34.3 of TS 23.222 [2]. 5. With reference to step 5 in clause 8.34.3 of TS 23.222 [2], the response in an OAuth 2.0 access token response. The access token includes the API invoker ID, resource owner ID, and the authorized scope of access. 6. With reference to step 6 in clause 8.34.3 of TS 23.222 [2], the request includes the received access token in step 5. The request is sent over a secure connection on the CAPIF-2e reference point. 7. With reference to step 7 in clause 8.34.3 of TS 23.222 [2], AEF checks the request against the token, including: 1) checking the token integrity and 2) checking whether the resource in the API invocation request is compliant with the resOwnerId claim in the access token. Editor’s Note: clarification on proposed security enhancement is FFS.
837a5602c57c068290b1254827b20390
33.700-23
6.2.3 Evaluation
TBD
837a5602c57c068290b1254827b20390
33.700-23
6.3 Solution #3: Client credentials flow based group authorization
837a5602c57c068290b1254827b20390
33.700-23
6.3.1 Introduction
This solution addresses KI#1. This solution uses the client credentials flow to enable the group authorization. Specifically, the CCF uses the locally stored group related authorization information to authorize the API invoker.
837a5602c57c068290b1254827b20390
33.700-23
6.3.2 Solution details
Figure 6.3.2-1: Client credentials flow based group authorization It is assumed that the group resource owner has provisioned the group authorization information to the CCF. 1-3. are identical to steps 1-3 defined in clause 8.34.3 of TS 23.222 [2]. 4. The CCF identifies the group authorization information based on the group identifier. The CCF authorizes the API invoker based on the group authorization information. 5. is similar to step 5 defined in clause 8.34.3 of TS 23.222 [2]. The authorization response includes the token. The token additionally includes the group identifier, which is used to indicate that the token is generated with the group authorization information. Editor’s Note: The group identifier in the token is FFS. 6-7. are identical to steps 6-7 defined in clause 8.34.3 of TS 23.222 [2].
837a5602c57c068290b1254827b20390
33.700-23
6.3.3 Evaluation
Editor’s Note: Evaluation is FFS.
837a5602c57c068290b1254827b20390
33.700-23
6.4 Solution #4: Supporting Group Authorization based on authorization information provided by GRO
837a5602c57c068290b1254827b20390
33.700-23
6.4.1 Introduction
This solution aims to address KI#1 to support authorization of a UE-hosted API invoker accessing resources owned by other UEs that belong to the same group. The solution proposes to reuse the TS 33.122 [x] clause 6.5.3 with the following enhancement: 1) Authorization information provided by GRO(for simplicity called GRO authorization information) additionally includes the group identifier and a description of which UEs’ resources within a group the API invoker on one UE can access. CCF uses the group identifier in the GRO authorization information (assumed to be GID1) and the group identifier received from the API invoker to find the correct authorization information, i.e., the GRO authorization information identified by GID1. 2) CCF authorizes the API invoker based on GRO authorization information locally available.
837a5602c57c068290b1254827b20390
33.700-23
6.4.2 Solution details
The authorization information provided by GRO(for simplicity called GRO authorization information) is transferred between the ROF and the CCF via the secure CAPIF-8 reference point. Editor’s Note: the communication between ROF and CCF is FFS. The GRO authorization information contains the same information of authorization information specified in TS 33.122 6.5.3.1, as well as the group identifier and a description of which UEs’ resources within a group the API invoker on one UE can access. CCF uses the group identifier in the GRO authorization information (assumed to be GID1) and the group identifier received from the API invoker to find the correct authorization information, i.e., the GRO authorization information identified by GID1. If using oauth client credential flow, the CCF checks whether the API invoker deployed in UE-2 is entitled to consume the API and allowed to access the resources of UE-1 of the same group based on GRO authorization information locally available. If using authorization code (optional PKCE) flow, the CCF checks whether the API invoker deployed in UE-2 is entitled to consume the API and allowed to access the resources of UE-1 of the same group based on GRO authorization information locally available at the execution time of issuing the authorization code.
837a5602c57c068290b1254827b20390
33.700-23
6.4.3 Evaluation
TBD
837a5602c57c068290b1254827b20390
33.700-23
6.5 Solution #5: Group authorization for UE-deployed API invoker accessing other UEs' resources of a group
837a5602c57c068290b1254827b20390
33.700-23
6.5.1 Introduction
This solution addresses the security requirements of Key issue#1. It is proposed to use the procedure as specified in clause 8.24 of TS 23.222 [2] and include group identifier as an optional parameter in the access token.
837a5602c57c068290b1254827b20390
33.700-23
6.5.2 Solution details
Figure 6.5.2-1: Procedure for Group Authorization for UE-deployed API invoker accessing other UEs' resources of a group 1. The API invoker (e.g., in UE 2) sends an Obtain service API authorization request to the CCF for obtaining permission to access the service API for other UE's resources hosted in the network (e.g., location). The request includes API invoker information, the group identifier, the UE in a group whose resources are to be accessed, scope information, and the identity of UE2. 2. CCF performs authentication of the API invoker (using authentication information) as specified in 3GPP TS 33.122 [3]. 3. The CCF, based on the group identifier and resource owner ID determines the identity of the GRO responsible for the group of UEs. 4. CCF performs the resource owner authorization check using the GRO as the RO for the requested resources of other UE(s) belonging to the group. Editor’s Note: How CCF reaches group resource owner is FFS. 5. Based on the successful group resource owner authorization, the CCF provides an access token that includes the resource owner ID, group identifier (optional), API invoker information and scope information. Editor’s Note: Which resource owner ID is included in access token is FFS. Editor’s Note: Whether addition of group identifier in access token is enough for authorization is FFS. 6. The API invoker sends service API invocation request to the API exposing function with the RO authorization information.
837a5602c57c068290b1254827b20390
33.700-23
6.5.3 Evaluation
TBD
837a5602c57c068290b1254827b20390
33.700-23
6.6 Solution #6: Addressing security of Open Discovery Service API
837a5602c57c068290b1254827b20390
33.700-23
6.6.1 Introduction
Open service API introduces the possibility for a requestor of accessing non-sensitive API Information before on-boarding. Due to the publicity of the information, i.e., non-sensitive information, there is no need to authorize the requestor at CCF. To ensure the correctness of the information provided by CCF to the requestor and the security of the communication, TLS should be used between the two entities.
837a5602c57c068290b1254827b20390
33.700-23
6.6.2 Solution details
1. Requestor will initiate a TLS connection with server-side certificate verification, towards CCF. 2. Requestor initiates the open discovery service API request with CCF and retrieves the required information as detailed in TS 23.222 [2].
837a5602c57c068290b1254827b20390
33.700-23
6.6.3 Evaluation
Editor’s Note: Evaluation is ffs.
837a5602c57c068290b1254827b20390
33.700-23
6.7 Solution #7: Security procedure for open discover service APIs
837a5602c57c068290b1254827b20390
33.700-23
6.7.1 Introduction
This solution addresses key issue #2 (Security for open discover service API). Open Discover Service APIs procedure introduced in TS 23.222 allows API invokers not recognized by the CAPIF Core Function to discover APIs without being onboarded to the CAPIF Core Function.
837a5602c57c068290b1254827b20390
33.700-23
6.7.2 Solution details
The requester who wants to discover service API information about the available set of APIs offered by CCF before onboarding and the CCF who supports open discover service APIs follows the procedure explained below for security of the open discover service APIs procedure specified in clause 8.38 of TS 23.222 [3]. The security information flow is depicted in Figure 6.7.2-1. It is assumed the Requestor has a discovery credential (e.g., an access token provided by the API provider domain), the address of the Open Discovery API of the CCF and optionally the root CA certificate of the CCF (e.g. provided by the API provider domain). The format and content of the discovery credential is not in the scope of the solution and depends on the agreement between CAPIF provider domain and API provider domain. The discovery credential can include authorization data which allows the Requester to obtain information about the APIs of the API provider domain, identifier of the Requester, and identifier of the API provider domain (e.g., API publisher information). For example, if the discovery credential is an OAuth 2.0 access token, then the identifier of the Requester can be included in "client_id" claim and the identifier of the API provider domain can be included in the issuer ("iss") claim as specified in RFC 7519 [6]. The CCF can also hold the authorization data provided by the API provider domain with a mechanism not specified in this solution. The authorization data obtained by using discovery credential or obtained from the API provider domain can indicate which requesters are authorized for open discover service APIs and for which APIs (e.g., any requestor or a list of requesters for any APIs or a list of APIs). NOTE: The OAuth 2.0 access token is given as an example for onboarding credential in the API invoker onboarding procedure (clause 6.1). That access token can also include authorization information for the open discover service APIs. In that case, same access token can be used for both open service API discover request and onboarding of the API invoker. In the steps below, "discovery credential" is used but if onboarding credential with the mentioned extension is used then "discovery credential" is replaced with "onboarding and discovery credential". Figure 6.7.2-1: Open discover service APIs security procedure 1. The requester establishes a TLS channel with the CCF and authenticates the CCF based on the CCF certificate. The requestor sends the discovery credential to the CCF. 2. The CCF verifies the discovery credential and obtains the identifier of the Requester and the identifier of the API provider domain. The CCF obtains the open discover service APIs related authorization data from the discovery credential or from the provisioned data available at the CCF. Then, the CCF performs filtering of service APIs information by taking the authorization data, the identifier of the Requester if available and the identifier of the API provider domain if available into account. 3. If the verification of the discovery credential is successful, the CCF returns the filtered service API information to the Requestor, otherwise returns an error. Editor’s Note: Whether OAuth token can be used for Authentication is FFS.
837a5602c57c068290b1254827b20390
33.700-23
6.7.3 Evaluation
Editor's Note: Evaluation is FFS.
837a5602c57c068290b1254827b20390
33.700-23
6.8 Solution #8: TLS based secure open service API discover
837a5602c57c068290b1254827b20390
33.700-23
6.8.1 Introduction
This solution addresses KI#2. Specifically, the TLS is used to protect the open service API discover procedure. The CCF’s local policy is used for requestor authorization.6.8.2 Solution details Figure 6.8.2-1: Open Discover service APIs 0. It is assumed that the requestor is preconfigured the certificate chain used to verify the CCF’s certificate. The requestor authenticates the CCF based on the CCF’s certificate. Then the UE builds TLS based on CCF’s certificate. Thus, the messages exchanged between UE and CCF are confidentiality, integrity, and replay protected from unauthorized parties. 1. is identical to step 1 defined in clause 8.38.3 of TS 23.222 [2]. 2. is similar to step 2 defined in clause 8.38.3 of TS 23.222 [2]. The only change is given as follows. With local policy, the CAPIF core function performs filtering of service APIs information. Editor’s Note: The local policy is FFS. 3. is identical to step 3 defined in clause 8.38.3 of TS 23.222 [2].
837a5602c57c068290b1254827b20390
33.700-23
6.8.3 Evaluation
Editor’s Note: Evaluation is FFS.
837a5602c57c068290b1254827b20390
33.700-23
6.9 Solution #9: Augmenting scope parameter with purpose information
837a5602c57c068290b1254827b20390
33.700-23
6.9.1 Introduction
This solution addresses key issue #3 and consists of augmenting the scope parameter in the token/authorization request and the token with purpose information; the resource owner authorization revocation request would likewise include purpose information. As aligned with the key issue description, the solution is not aiming to specify different purpose values, but to describe the usage of purpose information in authorization. NOTE: An example of purpose definition is found in W3C Data Privacy Vocabulary.
837a5602c57c068290b1254827b20390
33.700-23
6.9.2 Solution details
- For the client credential flow in CAPIF RNAA, the purpose information is included in the scope parameter of the token request. - For the authorization code flow in CAPIF RNAA, the purpose information is included in the scope parameter of authorization request. - The scope parameter in the issued token includes the purpose information. NOTE: How to encode the purpose information into the scope parameter of the requests and of the token is not in the scope of this solution. - In the revocation procedure, it is proposed that the purpose information can also be sent to the CCF and the CCF can use this information in identification of CAPIF RNAA tokens to be revoked. This solution proposes to leave how to structure the resource owner authorization revocation request to the implementation because the details of CAPIF-8 reference point is out of scope of 3GPP. Editor’s Note: Weather the purpose of information is needed in the request is FFS.
837a5602c57c068290b1254827b20390
33.700-23
6.9.3 Evaluation
Editor's Note: Evaluation is FFS.
837a5602c57c068290b1254827b20390
33.700-23
6.10 Solution #10: Purpose based authorization and authorization revocation
837a5602c57c068290b1254827b20390
33.700-23
6.10.1 Introduction
This solution addresses the KI#3. Specifically, if API invoker needs to obtain resource owner’s data from the network, the data processing purpose is used to determine whether CCF issues the token to the API invoker.
837a5602c57c068290b1254827b20390
33.700-23
6.10.2 Solution details
For RNAA related client credentials flow, the following enhancements are needed to support the purpose based authorization. • If the API invoker needs to get resource owner’s data from the network, the API invoker sends the data processing purpose (e.g., the location data is used for advertising) to the CCF. • The CCF authorizes the API invoker based on authorization information. The authorization indicates whether the data processing purpose is authorized. If the data processing purpose is not authorized, the CCF terminates the authorization procedure and will not send the token to the API invoker. For authorization using authorization code (optional PKCE) flow, the following enhancements are needed. • If the API invoker needs to get resource owner’s data from the network, the API invoker sends the data processing purpose (e.g., the location data is used for advertising) to the CCF via the ROF. • The CCF authorizes the API invoker based on authorization information provided by the ROF. The authorization indicates whether the purpose is authorized. If the data processing purpose is not authorized, the CCF terminates the authorization procedure and will not send the token to the API invoker. For authorization information and authorization revocation information transferring part, the following enhancement is needed. • ROF sends the allowed/disallowed data processing purpose to the CCF via CAPIF-8. Editor’s Note: Purpose delivery via CAPIF-8 is FFS.
837a5602c57c068290b1254827b20390
33.700-23
6.10.3 Evaluation
Editor’s Note: Evaluation is FFS.
837a5602c57c068290b1254827b20390
33.700-23
6.11 Solution #11: Enhancing finer granularity for purpose of information
837a5602c57c068290b1254827b20390
33.700-23
6.11.1 Introduction
This solution is addressing KI#3 by enhancing authorization mechanism to validate the purpose for retrieving the information. The solution proposes to enhance the already existing mechanisms available in CAPIF ecosystems, i.e., the access token as part of RNAA procedure. After authentication between the CCF and the API Invoker, the latter will include the required additional information to CCF during the Access token Request. The API Invoker will include in the scope parameter more authorization details that allow to distinguish. When the verification is completed, the CCF will include the authorization details, together with the purpose, into the access token returned to the API Invoker. The previously provided access token will allow the AEF to correctly authorize, or deny, the request by enhancing the mechanism already available to AEF.
837a5602c57c068290b1254827b20390
33.700-23
6.11.2 Solution details
6.11.2.1 Authorization provisioning Pre-requisites: CAPIF-1e authentication and secure session establishment is performed as specified in subclause 6.3.1 of 33.122. 1. After successful establishment of TLS session over CAPIF-1e, the API invoker shall send an Access Token Request message to the CAPIF core function with the new details, i.e., including the purpose of the request. 2. The CAPIF core function shall verify the Access Token Request message, by checking the allowed purpose for the specific API Invoker. NOTE: it is assumed that CCF is aware of the purposes that the various API will support. 3. The CCF will generate the access token including the purposes for which the API Invoker is allowed to request the data. 4. After establishing the secure session with the AEF, the API Invoker will send the service request to the AEF by including the purpose of requesting the resources. Editor’s Note: whether the purpose of the request should also be included in the request is ffs. 5. 6. In addition to traditional checks, AEF will also verify that the purposes included in the token are the same of the one in the service request. After successful authorization, the AEF will reply with the requested information. 6.11.2.2 Revocation procedure Release 19 defined the procedure to revoke RNAA token. Reusing the same procedure, i.e., sending the access token as part of the Revoke_Authorization service operation, will allow to revoke the purpose specific token.
837a5602c57c068290b1254827b20390
33.700-23
6.11.3 Evaluation
6.Y Solution #Y: <Title> 6.Y.1 Introduction 6.Y.2 Solution details 6.Y.3 Evaluation
837a5602c57c068290b1254827b20390
33.700-23
7 Conclusions
Annex <X>: Change history Change history Date Meeting TDoc CR Rev Cat Subject/Comment New version 2025-10 SA3#124 S3-253327 Skeleton 0.0.0 2025-10 SA3#124 S3-253731 Incorporate pCRs that add S3‑253756, S3‑253757, S3‑253758, S3‑253759, S3‑253761, S3-253760 0.1.0 2025-11 SA3#125 S3-254212 Incorporate pCRs that add S3-254176, S3‑254594, S3-254595, S3-254596, S3-254722, S3-254723, S3-254597, S3‑254598, S3-254599, S3-254600, S3-254601, S3-254062 0.2.0
ca180014f5532ca12c04dc24c74a1668
33.714
1 Scope
Editor’s Note: This clause is going to capture the scope of this study. This present document aims to identify potential threats and security requirements to support additional features for AIoT in Rel-20. Specifically, • Security aspects of concluding on authorization of intermediate UE for AIoT services in Topology 2 Editor’s note: which types of AIoT device are in the scope of topology 2 is FFS. NOTE 1: AIoT device Type 1 is restricted to isolated private network. Editor’s note: The aspect outlined in NOTE 1 needs to be reflected in the AIoT phase 2 Study Item update. • Security aspects to support DO-A Capable AIoT Devices • Identifies potential threats and new security requirements • Security mechanisms to support DO-A type AIoT communications in order to fulfil the identified security requirements • Security aspects of the AIOT system for public networks • Applicability of security requirements and procedures developed in TS 33.369 for isolated private networks will be re-assessed for Rel-20 AIoT system for public network NOTE 2: For AIoT device credentials storage and processing in public networks, the guidance in SP-250852 will be followed. Editor’s Note: Guidance given in SP-250852 will be copy pasted in NOTE 2 to replace the reference to SP-250852. NOTE 3: SNPN will follow the requirements for credentials storage and processing in TS 33.369.
ca180014f5532ca12c04dc24c74a1668
33.714
2 References
The following documents contain provisions which, through reference in this text, constitute provisions of the present document. - References are either specific (identified by date of publication, edition number, version number, etc.) or non‑specific. - For a specific reference, subsequent revisions do not apply. - For a non-specific reference, the latest version applies. In the case of a reference to a 3GPP document (including a GSM document), a non-specific reference implicitly refers to the latest version of that document in the same Release as the present document. [1] 3GPP TR 21.905: "Vocabulary for 3GPP Specifications". [2] 3GPP TR 23-700-13: "Study on Architecture Support of Ambient power-enabled Internet of Things". [3] 3GPP TR 38.848: "Technical Specification Group Radio Access Network; Study on Ambient IoT (Internet of Things) in RAN". [4] 3GPP TR 23700-30: "Study on Architecture support of Ambient power-enabled Internet of Things (AIoT); Phase 2". [5] 3GPP TR 38.769: "Study on solutions for Ambient IoT (Internet of Things) in NR". [6] 3GPP TS 22.369: "Service Requirements for ambient power-enabled IoT". [7] 3GPP TS 23.369: "Architecture support for Ambient power-enabled Internet of Things; Stage 2". [8] 3GPP TS 33.369: "Security aspects of Ambient Internet of Things (AIoT) services for isolated private networks". [9] 3GPP TS 33.501: "Security architecture and procedures for 5G System".
ca180014f5532ca12c04dc24c74a1668
33.714
3 Definitions of terms, symbols and abbreviations
ca180014f5532ca12c04dc24c74a1668
33.714
3.1 Terms
For the purposes of the present document, the terms given in TR 21.905 [1] and the following apply. A term defined in the present document takes precedence over the definition of the same term, if any, in TR 21.905 [1]. example: text used to clarify abstract rules by applying them literally.
ca180014f5532ca12c04dc24c74a1668
33.714
3.2 Symbols
For the purposes of the present document, the following symbols apply: <symbol> <Explanation>
ca180014f5532ca12c04dc24c74a1668
33.714
3.3 Abbreviations
For the purposes of the present document, the abbreviations given in TR 21.905 [1] and the following apply. An abbreviation defined in the present document takes precedence over the definition of the same abbreviation, if any, in TR 21.905 [1]. <ABBREVIATION> <Expansion>
ca180014f5532ca12c04dc24c74a1668
33.714
4 Key issues
Editor’s Note: This clause contains all the key issues identified during the study.
ca180014f5532ca12c04dc24c74a1668
33.714
4.1 Key Issue #1: Authorization of intermediate UE for 5G Ambient IoT services
ca180014f5532ca12c04dc24c74a1668
33.714
4.1.1 Key issue details
In TR 23.700-13 [2], Key Issues #1 and #3 describe the issues on the system architecture and procedure to support 5G Ambient IoT services, furthermore TR 23.700-30 [4], KI#1 describes the issues on the support AIoT services under the RRC-based option for UE Reader connectivity. The architecture for topology 2 is defined in TR 23.700-13 [2] clause 8.1.3 which forms the baseline for the release 20. In the Topology 2 as defined in TR 38.848 [3], the UE is acting as the intermediate node responsible for transferring the information between AIoT device and 5GS. If the authorization and authentication of the intermediate node is not supported, the attacker can play the role of an intermediate node and arbitrarily deny 5G AIoT service to the AIoT device. Therefore, it is necessary to study how to authorize a UE for acting as the intermediate node i.e an AIoT reader.
ca180014f5532ca12c04dc24c74a1668
33.714
4.1.2 Security threats
If the 5GC do not authorize the UE acting as an intermediate node, the attacker UE may misuse the Ambient IoT services provided by the core and hereby impersonate an authorised intermediate node. Editor’s Note: The threats may be refined based on SA2 agreed procedures.
ca180014f5532ca12c04dc24c74a1668
33.714
4.1.3 Potential security requirements
The 5GS shall be able to support the authorization of the AIoT capable UE as an intermediate node. Editor’s Note: Requirements are FFS.
ca180014f5532ca12c04dc24c74a1668
33.714
4.2 Key Issue #2: Authentication for AIoT devices
ca180014f5532ca12c04dc24c74a1668
33.714
4.2.1 Key issue details
DO-A capable AIOT devices can inform the network of their presence and send data to the AIOTF autonomously. The TR 23.700-30 [4] studies the architecture framework and procedure for DO-A capable AIoT devices, including the device initiated registration-like procedure and data transfer procedure. With the capability of providing information autonomously, the existing security mechanisms (e.g. authentication procedure) specified for DT capable AIoT devices need be enhanced to accommodate DO-A use cases. The authentication between the DO-A capable AIoT device and the network is required upon device-initiated communication to validate each other’s identities. Otherwise, the attacker may impersonate the victim device and send fake identification to the network side. Therefore, it is necessary to study how to perform authentication between the AIoT device and network, addressing risks such as impersonation. NOTE 1: For AIoT device credentials storage and processing in public networks, the AIoT device credentials storage will use UICC. The exact form factor of UICC, i.e. whether it is removable, non-removable or integrated is out of scope of 3GPP.
ca180014f5532ca12c04dc24c74a1668
33.714
4.2.2 Security threats
An attacker may impersonate the victim AIoT device and report fake identification to the network side. If the billing is based on per AIoT device’s identity, the fake identity may lead to charging problem. This can be used by an adversary to steal an AIoT device by replacing the AIoT device with a fake device, which might cause a loss to the owner of the device. An attacker can impersonate a legitimate network and communicate with AIoT device.
ca180014f5532ca12c04dc24c74a1668
33.714
4.2.3 Potential security requirements
The 5G system shall provide a means to perform mutual authentication between the DO-A capable AIoT device and the network. NOTE 2: AIoT device Type 1 is restricted to isolated private network. Editor’s Note: The aspect outlined in NOTE 2 needs to be reflected in the AIoT phase 2 Study Item update.
ca180014f5532ca12c04dc24c74a1668
33.714
4.3 Key Issue #3: Protection of information to support DO-A Capable AIoT Devices during AIoT service communication
ca180014f5532ca12c04dc24c74a1668
33.714
4.3.1 Key issue details
As per TS 22.369 [6], Ambient power-enabled IoT (AIoT) services aim to support various use cases, including inventory taking, sensor data collection, asset tracking, and actuator control. These services intended to operate with lower power consumption and complexity than the existing IoT technologies such as eMTC, NB-IoT, and RedCap. To fulfil these requirements, AIoT devices require a communication capability. From a security perspective, security mechanisms to protect the information transmitted during AIoT service communication need to be supported. Failure to provide such security mechanisms will lead to various attacks such as eavesdropping, manipulation and/or unauthorized transmission of the information during AIoT service communication.
ca180014f5532ca12c04dc24c74a1668
33.714
4.3.2 Security threats
In addition to the command operation (e.g., write, read) as specified in TS 23.369 [7], DO-A Capable AIoT Device can send data to the AIOTF autonomously. The following threats are still applicable: An attacker may acquire data transmitted to/from AIoT devices by eavesdropping messages if the communication of AIoT service is not confidentiality protected. An attacker may manipulate information during communication of AIoT service if the communication of AIoT service is not integrity protected. An attacker may replay a message if replay protection is not activated.
ca180014f5532ca12c04dc24c74a1668
33.714
4.3.3 Potential security requirements
The 5G system shall support a means to ensure confidentiality, integrity and/or replay protection of information transmitted between DO-A Capable AIoT Device and the network.
ca180014f5532ca12c04dc24c74a1668
33.714
4.4 Key Issue #4: DO-A capable AIOT device ID protection
ca180014f5532ca12c04dc24c74a1668
33.714
4.4.1 Key issue details
For AIoT device type 1, all communications between the network and the AIOT device are initiated by the network. Unlike AIOT device type 1, the DO-A AIOT device could autonomously initiate communication by sending a message to the network. Due to this change, privacy mechanisms specified in TS 33.369[8] for AIOT device type 1 may not be feasible for DO-A AIOT devices. Therefore, mechanisms for privacy of device ID of DO-A AIOT device contained in the message(s) exchanged between the device and the network should be studied.
ca180014f5532ca12c04dc24c74a1668
33.714
4.4.2 Threats
An attacker can identify, monitor and track a DO-A AIoT devices based on the identifiers associated with the AIoT device if the identifiers are not privacy protected.
ca180014f5532ca12c04dc24c74a1668
33.714
4.4.3 Potential security requirements
The 5G system shall support mechanisms to prevent privacy threats (e.g., identifying, linking, and tracking) against the identifier of the DO-A capable AIOT device(s).
ca180014f5532ca12c04dc24c74a1668
33.714
4.5 Key Issue #5: Amplification of resource exhaustion by exploiting AIoT paging messages
ca180014f5532ca12c04dc24c74a1668
33.714
4.5.1 Key issue details
Paging of AIoT devices is different than "regular" paging of regular UEs. In AIOT, one single paging message coming from the reader/network can be used to trigger multiple devices to respond by using, for example, a mask/filter based on target device identification, or by a group ID of the target devices. Once the target devices are triggered, the reader, core network of the PLMN, and the associated AF participate in various steps to accomplish the intended tasks, e.g., inventory reporting and command executing. Unlike regular paging, AIOT paging can happen for devices that are not necessarily already registered in the core network and hence cannot share a session security context with the network. The paging message can include information that the devices and core network of the PLMN can use in successful accomplishment of these tasks in those steps. Therefore, if parts of or the whole paging message is corrupted, the core network of the PLMN and the AF can end up wasting computational resources that leads to no successful accomplishment of the intended tasks. Moreover, the corrupted paging message results in waste of radio resources being used by AIOT over the air interface as well. The above can be used by an adversary that intentionally corrupt the paging message in a way so that many legitimate AIOT devices are triggered by the corrupted paging message, but later, in the core network of the PLMN or in the AF, the responses from the AIOT devices are found invalid. This happens not because the devices computed wrong responses, but because the devices used corrupted paging message in computing their responses. Such an attack can cause the PLMN and the AF wasting computational resources. It also causes the AIOT reader wasting radio resources that can adversely impact the regular UEs in the same network. If devices respond to a corrupted paging message, that should be identified as early as possible, and the responses should not be forwarded any further to the core network or to the AF.
ca180014f5532ca12c04dc24c74a1668
33.714
4.5.2 Security threats
An adversary can cause the core network of a PLMN or the AF wasting computational resources by corrupting or spoofing one single paging message, which is surprisingly little work on the adversary’s behalf, that triggers a lot of devices to send a paging response to the legitimate reader. The above attack can also cause the AIOT reader and serving NG-RAN node wasting radio resources that can adversely impact the regular UEs in the same network.
ca180014f5532ca12c04dc24c74a1668
33.714
4.5.3 Potential security requirements
Editor’s Note: Potential security requirements are FFS
ca180014f5532ca12c04dc24c74a1668
33.714
5 Solutions
Editor’s Note: This clause contains the proposed solutions addressing the identified key issues.
ca180014f5532ca12c04dc24c74a1668
33.714
5.0 Mapping of solutions to key issues
Editor’s Note: This clause captures mapping between key issues and solutions. Table 5.1-1: Mapping of solutions to key issues Key Issues Solutions 1 2 3 4 5 1 X 2 X 3 X 4 X 5 X 6 X 7 X 8 X
ca180014f5532ca12c04dc24c74a1668
33.714
5.1 Solution #1: Information protection after registration
ca180014f5532ca12c04dc24c74a1668
33.714
5.1.1 Introduction
This solution addresses KI#3.
ca180014f5532ca12c04dc24c74a1668
33.714
5.1.2 Solution details
ca180014f5532ca12c04dc24c74a1668
33.714
5.1.2.1 Registration procedure
The following figure depicts the AIoT registration procedure to activate information protection. Figure 5.1.2.1-1: AIoT registration procedure 1. AIoT device sends initial Register Request (Device ID, Device security capabilities) towards AIoTF. Editor’s Note: How to protect security capabilities is ffs. Editor's Note: Alignment with SA2 is FFS. NOTE: Protection of device ID is out of this solution. 2. AIoTF sends Authentication Request (Device ID) towards ADM. 3. AIoT device, AIoTF and ADM performs authentication procedure proposed for KI#2, e.g., using 5G AKA or EAP-AKA'. 4. After successful authentication, AIOTF derives KCOMM_ENT and KCOMM_INT from agreed key during authentication procedure (e.g., KAIOTF) for the AIoT device. AIoT device derives KCOMM_INT and KCOMM_ENC from KAIOTF same way as AIOTF does. Editor’s Note: Clarification on KAIOTF derivation on AIoT device is ffs. 5. AIOTF selects integrity and confidentiality algorithm based on Device security capabilities and algorithm priority list. The AIOTF generates the Register Response and integrity protects the Register Response with the KCOMM_INT and selected integrity algorithm, then partially encrypts the Registration Response with the KCOMM_ENC and selected confidentiality algorithm with the selected algorithms in clear text. The AIOTF sends the protected Register Response to AIoT device, and starts ciphering/deciphering (i.e., the Registration Response also is for NAS Security Mode Command). AIoT device integrity checks the protected Register Request, and if successful, decrypts the protected Register Response then starts ciphering/deciphering and integrity protection.
ca180014f5532ca12c04dc24c74a1668
33.714
5.1.3 Evaluation
Editor’s Note: Each solution should motivate how the potential security requirements of the key issues being addressed are fulfilled. TBD
ca180014f5532ca12c04dc24c74a1668
33.714
5.2 Solution #2: Protection of information during AIoT service communication
ca180014f5532ca12c04dc24c74a1668
33.714
5.2.1 Introduction
This solution addresses key issue#3 on protection of information during AIoT service communication. The solution reuses the security mechanisms for NAS protection from TS 33.501 [9] modulo some simplifications in order to avoid the need for an additional security activation procedure. By comparison to the mechanisms specified in TS 33.369 [8], the solution introduces a security context and the counters for replay protection. This is because the solution assumes that the device may receive multiple successive commands after authentication. The solution assumes also that the AIOTF is the termination point for information protection.
ca180014f5532ca12c04dc24c74a1668
33.714
5.2.2 Solution details
It is assumed that following a successful authentication procedure, the device and the network derive a session key called KAIOTF, for example in a similar manner to the procedure in TS 33.369 [8]. The device stores this key as part of the security context until a new authentication run. The authentication procedure is not covered in this solution and is left to other solutions addressing key issue #2. Editor's Note: Unlike Rel-19, whether AIOTF has to maintain security contexts is FFS. In addition to the session key, both the device and the network maintain a pair of downlink and uplink counters for replay protection similarly to the NAS COUNTs specified in TS 33.501 [9]. The counters are maintained and updated similarly to how it is done for the NAS COUNTs in TS 33.501 [9], i.e., following a successful integrity check. In order to avoid an additional round trip of message to agree on the security algorithms, the selected ciphering and integrity protection algorithms are indicated in the downlink NAS message (e.g., command request) by the network to the device. Editor’s Note: Whether the indications are included in every command or the first one is FFS. NOTE: The format of the indications for algorithm selection (e.g., one bit or several bits) can be decided accordingly when the algorithms are decided. In order to cater for a potential loss of the NAS response message (e.g., command response), the network keeps including the selected algorithm indications until the successful reception of a NAS response message, in which case the network stores the selected algorithms as part of the security context. For the protection algorithms, the solution assumes that the AIOTF and device supports one or several of the algorithms specified in Annex D of TS 33.501 [9]. The solution does not take a stand on which and how many algorithms are to be supported. The lower level security keys KCommand_enc and KCommand_int are derived from the session key based on the signalled algorithms and are stored as part of the security context both on the device and the network side. When the security context is available, the DO-A device sends protected uplink NAS message including the DO-A data. The security context can be used for protection of downlink NAS message as well.
ca180014f5532ca12c04dc24c74a1668
33.714
5.2.3 Evaluation
TBD
ca180014f5532ca12c04dc24c74a1668
33.714
5.3 Solution #3: Protecting information for DO-A communication
ca180014f5532ca12c04dc24c74a1668
33.714
5.3.1 Introduction
KI#3 describes the need to “support a means to ensure confidentiality, integrity and/or replay protection of information transmitted between DO-A Capable AIoT Device and the network.” This solution intends to fulfill this requirement. The solution makes the following assumption: - DO-A communication is a new procedure (e.g., not re-using the inventory-and-command procedure) - DO-A device does not need to register with the network. - DO-A device and the network have a pre-shared key (e.g., Kaiot_root or Kaiotf as a result of a previous authentication) - DO-A device and the network maintain two counters (one network counter and one device counter) that starts with 0 and are incremented every time DO-A device sends data In this solution, the AIoT DO-A device uses its pre-shared key to generate a set of protection keys, use the protections keys to encrypt and integrity the DO-A data before sending the data in a protected NAS container to the network.
ca180014f5532ca12c04dc24c74a1668
33.714
5.3.2 Solution Details
1. DO-A device has data to send and initiates an AIoT DO-a data transmission request to NG-RAN. 1a. NG-RAN sends AIoT DO-A data transmission to AIOTF. 1b. If AIOTF does not already have the DO-A AIoT device context (e.g., Kaiotf or device counter value), AIOTF and ADM performs an AIOTF key retrieval procedure for the DO-A AIoT device. AIOTF takes the current values of the device counter and network counter and derives set of keys Kdo-a_enc and Kdo-a_int using either the Kaiot_root or Kaiotf (if Kaiotf exists for example, due to a prior authentication procedure). Key derivation function can reuse Annex A.3 or A.4 of TS 33.369[8]. 2. DO-A device prepares for the data to be protected. DO-A device takes the current values of the device counter and network counter and derives set of keys Kdo-a_enc and Kdo-a_int using either the Kaiot_root or Kaiotf (if Kaiotf exists for example, due to a prior authentication procedure). Key derivation function can reuse Annex A.3 or A.4 of TS 33.369[8]. Furthermore, the DO-A device constructs a AIOT NAS Container and protect the message based on the Kdo-a_enc and Kdo-a_int and the confidentiality and integrity algorithms for the AIoT device that has been pre-configured. The AIOTF shall send the protected Command Request containing an indication on whether ciphering is activated to NG-RAN. DO-A device increments the locally kept device counter and network counter by 1. 3. The DO-A device sends a D2R message containing the protected AIOT NAS Container to the NG-RAN. 4.The NG-RAN forwards the protected AIOT NAS Container to the AIOTF. 5.The AIOTF processes the protected AIOT NAS Container. If the verification of integrity is successful, the AIOTF then deciphers the protected AIOT NAS Container if ciphering is activated. AIOTF increments the device counter and network counter by 1. 6. The AIOTF forwards the processed data to the AF. Editor’s Note: The need for the first message is FFS. Editor’s Note: How AIOTF looks up the key for the DO-A device is FFS.
ca180014f5532ca12c04dc24c74a1668
33.714
5.3.3 Evaluation
TBD.
ca180014f5532ca12c04dc24c74a1668
33.714
5.4 Solution #4: ID privacy based on stored type T-ID
ca180014f5532ca12c04dc24c74a1668
33.714
5.4.1 Introduction
This solution addresses KI#4. This solution proposes to reuse T-ID update method of release 19 as much as possible. In release 19, two T-ID types are defined: concealed and stored. This solution propose to reuse stored type T-ID and corresponding update method.
ca180014f5532ca12c04dc24c74a1668
33.714
5.4.2 Solution details
The following figure depicts the AIoT device ID protection based on stored type T-ID. Figure 5.4.2-1: AIoT Device ID protection based on stored type T-ID 1. AIoT device is preconfigured with initial T-ID (i.e. T-ID0), which can be derived from AIoT device permanent ID. The ADM also stores the initial T-ID for the AIoT device. After some registration procedures, the initial T-ID has been updated to T-IDn. Editor’s Note: How to pre-configure the initial T-ID is ffs. Editor’s Note: How the solution applies to the T-ID variants as defined in TS 33.369 is FFS.2. AIoT device sends initial Register Request (T-IDn) towards AIOTF. 3. AIoTF sends Authentication Request (T-IDn) towards ADM. 4. AIoT device, AIOTF and ADM performs authentication procedure. 5-7. If authentication succeeds, AIOTF retrieves new T-ID from ADM as described in step 10 of clause 5.4.3 in TS 33.369 [8]. 8. As described in steps 2 and 3 of clause 5.4.3 in TS 33.369 [8] with exception that the AIOTF sends Register Response with the T-ID handling information to AIoT device. Editor’s Note: Update on usage of the same T-ID between registration is ffs. 9. As described in "after step 10" of clause 5.4.3 in TS 33.369 [8]. The T-IDn+1 is used for successive initial registration procedure.
ca180014f5532ca12c04dc24c74a1668
33.714
5.4.3 Evaluation
Editor’s Note: Each solution should motivate how the potential security requirements of the key issues being addressed are fulfilled. Editor’s Note: Evaluation of T-ID storage on device side in de-registration state is ffs. TBD
ca180014f5532ca12c04dc24c74a1668
33.714
5.5 Solution #5: Privacy-preserving device identification responding to group paging using AICI
ca180014f5532ca12c04dc24c74a1668
33.714
5.5.1 Introduction
This solution addresses KI#4: AIOT device ID protection in DO-A procedure. The solution describes how a device identifies itself to the network in response to a group paging message, when the device does not have an established session or registered state with the network. The solution uses AIoT Concealed Device Identifier (AICI) generated by concealing the AIoT device’s long-term identifier. Compared to the procedure in TS 33.369, this solution is expected to reduce the computation overhead at the network, since the network does not need to compare the RES with the XRES for all devices of the group.The AICIs are pre-computed by the network using a public key of the network — the encryption algorithm to produce an AICI is randomized, i.e., each AICI is different even when the long-term identifier of the device is the same. The solution proposes that in response to a group paging message, the AIoT device sends a message to the network that includes an AICI. The solution proposes that the 5G network computes an AICI and provide the AICI to the AIoT device in a command message. Once a network authenticates an AIoT device, the network can send a command message to the AIoT device. In the command message, the network includes a new AICI, which is computed based on the long-term identifier of the AIoT device using the public key of the network. The downlink command message is both confidentiality and integrity protected using keys derived from the shared key KAIOT_root between the network and the AIoT device.
ca180014f5532ca12c04dc24c74a1668
33.714
5.5.2 Solution details
Figure 5.5.2-1 presents a high-level message flow of the solution. The figure is described step-by-step in the following: In Step 0, the ADM provides the AIOTF necessary information to page a group of devices — e.g., an identifier identifying a group (let us call it a group identifier) and an authentication challenge. NOTE 1: How the network creates a group identifier and how a device checks if the device belongs to the group identified by the group identifier is out of scope of this solution. Instead, this solution assumes a group identifier is used in group paging. In TS 33.369, the same purpose is served by using filtering information. In Step 1, the AIOTF sends a paging request to the AIoT reader/gNB by including the information necessary for group paging — for example, a group identifier and the authentication challenge RANDAIOT_n. In Step 2, the AIoT reader/gNB broadcasts a paging message that includes the group identifier and the authentication challenge RANDAIOT_n. In Step 3, The device checks if it is part of the group identified by the group identifier. If the device belongs to the group, it generates another authentication challenge RANDAIOT_d, then the device computes the authentication challenge response RES based on authentication challenges RANDAIOT_n and RANDAIOT_d, using the shared key KAIOT_root with the network. The device first checks if it has a network-provided AICI or not — if it does not have a network provided AICI, then it computes AICI using a null scheme. In Step 4, The device sends a response to the AIoT reader. The device includes an AICI, RANDAIOT_d, and RES in the response. In Step 5, the AIoT reader/gNB forwards the response received in Step 4 to AIOTF. In Step 6, the AIOTF forwards the message received in Step 5 to ADM. In Step 7, the ADM deconceals AICI into long-term identifier — using the private key corresponding to the public key the network used to conceal the AICI when it sent the AICI to the device earlier. Then the device checks, authentication challenges RANDAIOT_n and RANDAIOT_d using the shared key KAIOT_root for the device, wheter response to authentication challenge RES is valid. The ADM computes a new AICI´ using the key used for computing AICI (i.e., the public key of the network), and derives a session key KAIOTF from the shared key KAIOT_root for device to protect a downlink command message In Step 8, the ADM forwards the device’s long-term ID, AICI´ and KAIOTF to the AIoTF. In Step 9, the AIOTF generates two keys KCommand_enc and KCommand_int, and prepares a command message that includes AICI´, encrypts the command message using KCommand_enc and computes a MAC of the encrypted command message using the key KCommand_int. In Step 10, the AIOTF forwards the encrypted command message to the AIoT reader/gNB. In Step 11, the AIoT reader/gNB forwards the encrypted command message and the MAC to the AIOT device. In Step 12, the device derives key KAIOTF from the shared key KAIOT_root, and derives keys KCommand_enc and KCommand_int from KAIOTF in the same manner as in ADM, validates MAC using KCommand_int and decrypts command message using KCommand_enc, and updates AICI with AICI´ Figure 5.5.2-1: Procedure for using AICI in response to group paging and delivering an AICI to an AIoT Device Exact content of the messages exchanged, and details about authentication challenge, computing response RES to the challenge, and deriving the keys KCommand_enc and KCommand_int are not described because these details have to be adjusted with the authentication protocol that is finally agreed. NOTE 2: If the AIoT device does not have a network-computed AICI, for example, in the very first time of the device’s life cycle, then the AIoT device computes AICI using null scheme. This happens only in the beginning. To avoid using null scheme in the first time, a network can choose to provision every AIoT device with a network-computed AICI before they are handed out to their users. NOTE 3: AICI is not stored in the network. Instead, the network decrypts AICI. On the other hand, a device accepts an AICI only if it is computed by the legitimate home network — hence a device cannot obtain an AICI that the network won’t recognize. Therefore, any question about AICI synchronization is not relevant. NOTE 4: The solution requires AIoT devices to have the capability to update and store AICI.
ca180014f5532ca12c04dc24c74a1668
33.714
5.5.3 Evaluation
The solution assumes that group paging is used to page DO-A capable devices that are not registered to the network. Editor’s Note 1: Further Evaluation is FFS
ca180014f5532ca12c04dc24c74a1668
33.714
5.6 Solution #6: Privacy-preserving group paging using Bloom filter
ca180014f5532ca12c04dc24c74a1668
33.714
5.6.1 Introduction
This solution addresses KI#4: AIOT device ID protection in DO-A procedure. The solution uses a Bloom filter to page a group of devices. First the network generates privacy-preserving concealed identifiers for every device in the group to be paged. Then the network inserts the privacy preserving concealed identifiers into a Bloom filter, and sends the Bloom filter as a compressed paging identifier as part of the paging message.
ca180014f5532ca12c04dc24c74a1668
33.714
5.6.2 Solution details
The proposed solution is explained step-by-step in the following: Step1. The AIOTF sends group identification information (e.g., Filtering Information) to the ADM for the group that the AIOT is intends to reach to. Step2. the ADM computes a concealed temporary identifier CT-ID for each device in a group (let us call the group G) using an identifier ID of the device, a key KID associated with the device, and a freshness parameter RAND associated with the group using a hash function H: CT-ID = H(long-term device ID, K_AIOT_root, RAND) for each device in G It is noticeable that the freshness parameter RAND is not per device but remains the same for every device in the group G. The RAND can also be used as the authentication challenge to all the devices. Each time the ADM gnerates CT-IDs for a group of devices, the ADM chooses a fresh RAND. Step3. After computing all the CT-IDs, the ADM forwards the CT-IDs and the RAND to the AIOTF. Step4. The AIOTF inserts the CT-IDs in a Bloom filter B. A Bloom filter is a bit array of m bits and involves k hash functions (h1, …, hk) where each hash function hi has the range [0, m-1]. Insertion of CT-ID is done by setting the bit B[hi(CT-ID)] to 1 if it is not already set to 1, for all i in {1, … , k}. Step5. The AIOTF sends the Bloom filter B to the AIOT Reader/gNB in a paging request message by including the freshness parameter/authentication challenge RAND, the size m of the Bloom filter B and integer k representing the number of hash functions involved in the Bloom filter B. Step6. The AIOT Reader/gNB reader includes B, RAND, m, and k in a paging message and transmits the paging message over the air. Step7. Every AIOT device receiving the paging message computes their own CT-ID in the same manner ADM computed in Step 2. Then the device checks if the computed CT-ID is included in the Bloom filter B by using the parameter m an k. The CT-ID is considered available in the Bloom filter by checking whether B[hi(CT-ID)] is set to 1, for all i in {1, … , k} — if all those bits are set to 1, then the CT-ID is included in the Bloom filter, otherwise C-TID is not not included. If the CT-ID is found to be available in B, then the device considers that the paging message is meant for the device. Figure 5.6.2-1: Privacy-preserving group paging using Bloom filter. A Bloom filter is probabilistic, it cannot have any false neagatives but can have false positives, i.e., if a CT-ID is found to be not included in the Bloom filter, then it is truly not included. However, if a CT-ID is found to be included in the Bloom filter, then, with some measurable probability, it may be the case that the C-TID have actually never been inserted into the Bloom filter. Let us consider that the number of C-TIDs inserted into the Bloom filter is n. Then, the probability p of false positives is computable as a function of m, k, and n, where p ≈ (1 – e-kn/m)k. Therefore, the efficiency of a Bloom filter in the context of group paging in AIoT depends on what values of m, k, and n can be chosen in a practically feasible manner. Editor’s Note: Whether a Bloom filter can be constructed that can be accommodated within a constrained group paging message while maintaining acceptable false positive rate is FFS. Editor’s Note: Whether the devices have sufficient resources (e.g., power and energy) to compute k number of hash functions to process a Bloom filter suitable to use in AIOT group paging is FFS. Editor’s Note: Alignment of the solution with SA2-defined procedures is FFS.