hash stringlengths 32 32 | doc_id stringlengths 5 12 | section stringlengths 5 1.47k | content stringlengths 0 6.67M |
|---|---|---|---|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.10.2.3.2 Profile D (Hybrid 2)
| The ME and SIDF implement this profile. The parameters for this profile are the following:
- Identifier : SecP256r1MLKEM768 (Combining secp256r1 ECDH with ML-KEM-768)
- EC domain parameters : secp256r1 [10]
- EC Diffie-Hellman primitive : Elliptic Curve Cofactor Diffie-Hellman Primitive [9]
- point compression : true
- ML-KEM parameters : Level 3 (k, lattice dimension 3)
- KDF : ANSI-X9.63-KDF [9]
- Hash : SHA-256
- SharedInfo1 : (the ephemeral public key octet string– see [9] section 5.1.3)
- Shared secret key Z1 : Z (see [9] section 5.1.3) || Shared secret field from ML-KEM
- MAC : HMAC–SHA-256
- mackeylen : 32 octets (256 bits)
- maclen : 8 octets (64 bits)
- SharedInfo2 : the empty string
- ENC : AES–256 in CTR mode
- enckeylen : 32 octets (256 bits)
- icblen : 32 octets (256 bits)
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.10.3 Evaluation
| TBD
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.11 Solution #11 to SUCI calculation: SUPI Concealment using hybrid method
| Editor’s Note: Performances due to PQC operations performed after ECIES operations are FFS.
Editor’s Note: The pros and cons (including security, complexity and efficiency) of combining traditional asymmetric cryptographic algorithms with post-quantum cryptographic algorithms for SUCI calculation is FFS.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.11.1 Introduction
| Replacing classical cryptography with PQC algorithms at an early stage carries an inherent risk as a first time widespread deployment and more rigorous testing of PQC algorithms may be needed. So it will be beneficial to have it integrated with classical asymmetric cryptography based security mechanisms via a hybrid approach, where both classical asymmetric algorithms and post-quantum algorithms coexist. In case vulnerabilities are found in either type of algorithm, the presence of both classical and post-quantum algorithms in a hybrid setup reduces the impact of potential breaches, providing additional resilience to the overall cryptography. The hybrid method described here is applying PQC-based key encapsulation mechanism (KEM) to protect final output which is generated via ECIES.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.11.2 Solution details
| |
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.11.2.1 Processing on UE side
| The processing on UE side is done as follows.
Figure 7.2.1.11.2.1-1: SUCI generation using hybrid method at UE
1. UE generates a final output_ECC using ECIES as described in Annex C.3.2 in TS 33.501 [4], where the final output_ECC is Eph. EC public key||ciphertext||MAC tag.
2. UE generates an ephemeral shared key (KPQC) and an encrypted PQC shared key based on a PQC-based public key associated with the home network.
3. UE generates ephemeral symmetric encryption key and ephemeral MAC key using a KDF function and KPQC.
4. UE protects the final output_ECC using the encryption key and the MAC key. The final output is the concatenation of encrypted PQC shared key, ciphertext (i.e., Enc(Eph EC public key||ciphertext||MAC)), and MAC tag value.
Figure 7.2.1.11.2.1-1 defines the scheme output (i.e., the final output in step 4) as a result of the above steps, as defined in TS 23.003 [74].
Figure 7.2.1.11.2.1-2: Scheme output based on hybrid method
NOTE: Ciphertext output from PQC key encapsulation is referred to as encrypted PQC shared key as there is another ciphertext value from step 3 of symmetric encryption, to avoid confusion.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.11.2.2 Processing on home network side
| The processing on home network (HN) side is done as follows.
Figure 7.2.1.11.2-3: Decryption based on hybrid method at home network
1. Home network (HN) decapsulates the encrypted PQC shared key to derive the ephemeral shared key (KPQC).
2. HN generates ephemeral symmetric encryption key and ephemeral MAC key using a KDF function and KPQC.
3. HN verifies the MAC and decrypts the ciphertext to derive the final output_ECC, using the MAC key and encryption key respectively.
4. HN obtain the plaintext block (i.e., UE ID) using ECIES as described in Annex C.3.3 in TS 33.501 [4].
NOTE: Ciphertext input to PQC key decapsulation is referred to as encrypted PQC shared key as there is another ciphertext value from step 3 of symmetric decryption, to avoid confusion.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.11.3 Evaluation
| TBD
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.12 Solution #12 to SUCI calculation: Hybrid SUCI calculation
| 7.2.1.12.1 Introduction
This solution addresses the key issue#1.
7.2.1.12.2 Solution details
EN#1: The details of the Combiner Function (3>) is FFS.
SUPI and SUCI type
The SUPI is a globally unique 5G Subscription Permanent Identifier allocated to each subscriber in the 5G System. It is defined in clause 5.9.2 of 3GPP TS 23.501 [85].
The SUPI is defined as:
- a SUPI type: in this release of the specification, it may indicate an IMSI, a Network Specific Identifier (NSI), a Global Line Identifier (GLI) or a Global Cable Identifier (GCI); and
- dependent on the value of the SUPI type:
- an IMSI as defined in clause 2.1 of TS 23.003 [74];
- a Network Specific Identifier (NSI), taking the form of a Network Access Identifier (NAI) as defined in clause 28.7.2 of TS 23.003 [74];
- a Global Cable Identifier (GCI) taking the form of a NAI as defined in clause 28.15.2 of TS 23.003 [74];
- a Global Line Identifier (GLI) taking the form of an NAI as defined in clause 28.16.2 of TS 23.003 [74].
NOTE: Depending on the protocol used to convey the SUPI, the SUPI type can take different formats.
The SUCI is a privacy preserving identifier containing the concealed SUPI. It is defined in clause 6.12.2 of 3GPP TS 33.501 [4].
The SUCI is composed of the following parts:
1) SUPI Type, consisting in a value in the range 0 to 7. It identifies the type of the SUPI concealed in the SUCI. The following values are defined:
- 0: IMSI
- 1: Network Specific Identifier (NSI)
- 2: Global Line Identifier (GLI)
- 3: Global Cable Identifier (GCI)
- 4 to 7: spare values for future use.
2) Home Network Identifier, identifying the home network of the subscriber.
When the SUPI Type is an IMSI, the Home Network Identifier is composed of two parts:
- Mobile Country Code (MCC), consisting of three decimal digits. The MCC identifies uniquely the country of domicile of the mobile subscription.
- Mobile Network Code (MNC), consisting of two or three decimal digits. The MNC identifies the home PLMN or SNPN of the mobile subscription.
When the SUPI type is a Network Specific Identifier (NSI), a GLI or a GCI, the Home Network Identifier consists of a string of characters with a variable length representing a domain name as specified in clause 2.2 of IETF RFC 7542. For a GLI or a GCI, the domain name shall correspond to the realm part specified in the NAI format for SUPI in clauses 28.15.2 and 28.16.2.
3) Routing Indicator, consisting of 1 to 4 decimal digits assigned by the home network operator and provisioned in the USIM, that allow together with the Home Network Identifier to route network signalling with SUCI to AUSF and UDM instances capable to serve the subscriber.
Each decimal digit present in the Routing Indicator shall be regarded as meaningful (e.g., value "012" is not the same as value "12"). If no Routing Indicator is configured on the USIM, this data field shall be set to the value 0 (i.e., only consist of one decimal digit of "0").
4) Protection Scheme Identifier, consisting in a value in the range of 0 to 15 (see Annex C.1 of 3GPP TS 33.501 [4]). It represents the null scheme, or a non-null scheme specified in Annex C of 3GPP TS 33.501 [4], or a protection scheme specified by the HPLMN; the null scheme shall be used if the SUPI type is a GLI or GCI.
5) Home Network Public Key Identifier (traditional), consisting in a value in the range 0 to 255. It represents a public key provisioned by the HPLMN or SNPN and it is used to identify the key used for SUPI protection. This data field shall be set to the value 0 if and only if null protection scheme is used.
6) Home Network Public Key Identifier (PQC KEM), consisting in a value in the range 0 to 255. It represents a public key provisioned by the HPLMN or SNPN and it is used to identify the key used for SUPI protection apart from traditional HN public key. Example of the PQC KEM Public key is ML-KEM [21] (selected by NIST standards).
Note: If the above 6 needs to be avoided, then it could be merged with 5 and sent as a bitmap, where the bits are set for those traditional and PQC identifiers (known at USIM and UDM). Example: 01 for traditional alone, 10 for PQC alone and 11 for both traditional and PQC.
So, the bitmap and public keys ID (traditional& PQC) needs to fit into 5).
Moreover this 6) is optional and present only when hybrid key exchange is available.
7) Scheme Output, consisting of a string of characters with a variable length or hexadecimal digits, dependent on the used protection scheme, as defined below. It represents the output of a public key protection scheme specified in Annex C of 3GPP TS 33.501 [4] or the output of a protection scheme specified by the HPLMN.
PQC KEM ciphertext: Post Quantum Cryptography Key encapsulation mechanism uses the PQC KEM public key of Home Network to generate the ciphertext.
At UE, generate key pair (Ephemeral public key and private key) using key pair generation primitive. Based on the Diffie-Hellman primitive, a shared secret key element is derived (from public key of HN and generated ephemeral private key). PQC KEM ciphertext(ct) is generated using the Key encapsulation mechanism (asymmetric cryptographic scheme) where PK is PQC KEM public key of HN (PQC KEM public key is identified by PQC KEM public key ID and UDM/USIM has list of PQC KEM public keys). A PQC shared secret (ss) is also generated which is used as an input to a Key Derivation Function (KDF) to derive the final PQC shared secret.
Followed by that, key combiner function is used to combine the traditional shared key and newly generated final PQC shared secret (from the KDF function) to generate AEAD key K. With the derived key, symmetric encryption (AEAD) is performed to encrypt the plaintext block (SUPI) to generate the ciphered text and the MAC key.
This hybrid PQC solution is making use of two separate HN identifiers, one for the classic (traditional) and yet another for the hybrid PQC. Consequently, at SIDF, the received UE ephemeral public key and stored private key of home network (Traditional HN public key ID is used to fetch the corresponding HN private key) is used to generate the ephemeral shared key. PQC KEM ciphertext(ct) is used along with the PQC KEM secret key of HN– SK (PQC Public key ID received from UE is used to fetch the Secret Key SK in HN), in Key decapsulation mechanism (asymmetric cryptographic scheme) to generate the shared secret. The newly generated shared secret is used along with the traditional ephemeral shared key as inputs to the key combiner function functions to generate AEAD Key. The generated AEAD Key is used to de-cipher the cipher text using symmetric decryption (AEAD). The expected MAC is compared against the received MAC, and with this comparison the integrity of the SUCI is verified.
7.2.1.12.3 Evaluation
TBD
Editor’s Note: Further evaluation to be added.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.13 Solution #13 to SUCI calculation: Symmetric crypto based SUCI
| Overview: The UDM is creating a collection of relevant SUPI values. The UDM is encrypting by using symmetric crypto each of SUPI values from the list and is sending these to the UE. The UE is appending the selected encrypted SUPI and is hashing the encrypted SUPI together with the concatenated hashed Key KSUPI. The UDM can verify the authenticity of the SUPI. The UDM can verify the authenticity of the UE, and in successful case the UDM-UE auth can be processed.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.13.1 Introduction
| This solution addresses the key issue#1.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.13.2 Solution details
| Editor’s Note: The description of a resynchronisation procedure is FFS.
Editor’s Note: It is FFS about the first registration procedure.
Editor’s Note: It is FFS How are new encrypted SUPIs provisioned to UEs.
Editor’s Note: It is FFS How does the UDM regenerate the new symmetric keys for Pseudonyms.
Editor’s Note: Resynchronization of desynchronized RANDs is FFS.
Editor’s Note: Impact on fulfilling LI requirements is FFS.
Editor’s note: The elaboration for why the SUCI has to be encrypted is FFS.
Overview of Encrypted SUCI in UDM:
Figure 7.2.1.13.2-1: Encrypted SUCI in UDM (overview)
In the above Figure 7.2.1.13.2-1 the list of Encrypted SUCI is generated in UDM is shown. The UDM will first generate the key KSUPI using hash of SUPI, long term credentials of Subscriber K, RANDSUPI. Also, in parallel there will be list of Random numbers (RANDSUCI#1, RANDSUCI#2, etc) are generated using PRNG (Pseudo Random Number Generator). Use the newly generated key KSUPI to encrypt all the RANDSUCI#1, RANDSUCI#2, etc, to generate Enc(RANDSUCI#1), Enc(RANDSUCI#2), etc.
Overview of Temporary SUCI:
Figure 7.2.1.13.2-2: Temporary SUPI (overview)
The following steps are applicable:
Pre-configuration Phase
1.) The UDM is creating the list of SUCI values and is encrypting these by using symmetric cryptography. (Rationale: Actually, symmetric cryptography is assumed to be quantum safe.) The UDM is creating a Key KSUPI for the SUPI encryption and must store this key, because this key is later used for the decryption.
The SUCI values will be computed as shown by Figure 7.2.1.13.2-1. The SUCI values will be encrypted to provide privacy during transition phase.
The KSUPI is the hashed output value of input parameters/values (i.e., long-term key K, RANDSUPI, and hash value of SUPI). For the KDF, the hash functions of the SHA-3 family are considered quantum-resistant, i.e., digests (hash values) that are 128, 224, 256, 384 or 512 bits, are candidates for use in the KDF. It can be assumed, that the UDM has sufficient processing capacity to run the KDF for KSUPI computation.
2.) The list of encrypted SUCI’s is sent to the UE along with RANDSUPI. This RANDSUPI is used by UDM to generate the Key KSUPI (reference “Overview of Encrypted SUCI in UDM” of this document).
The RANDSUPI is a random value with predefined length and is used for freshness purposes. For random number generation the NIST Special Publication 800-90A [86] is to be used as reference. The RANDSUPI is not encrypted, which is similar to the RAND from the AV in EAP-AKA (refer to TS 33.501 [4]). The privacy of the UE can NOT be compromised by disclosing the RANDSUPI, because the RANDSUPI can NOT be used for identification of the subscriber.
3.) The UE is storing the received list of encrypted SUCI’s. USIM/ ME will also use RANDSUPI to generate Key KSUPI.
The storage place of the SUCI should non-volatile memory. RANDSUCI values have been used should be moved into volatile memory. The KSUPI must be processed, because this is providing a binding to the specific long-term key, basically, this is providing a proof-of-possession, i.e., the encrypted RANDSUCI, which will be sent to the Network includes the long-term key K.
Registration Phase
4.) The UE is now selecting one encrypted SUCI.
5.) The UE is sending the registration request to the UDM and is added new processed values into this message. The following needs to be processed by the UE: The root key is the Key K which is stored inside the USIM of that UE. The UE is creating a hash of that Key KSUPI. Furthermore, the UE is concatenating the Encrypted SUCI and is hashing both, the Encrypted SUCI together with the hashed Key KSUPI. The rationale for creating this concatenation is the following: The Encrypted SUCI is used by the UDM to verify the authentication of the SUCI value, while the hashed key KSUPI is used by the UDM to verify the authenticity of the UE (could also be called, the legitimacy of the UE for sending these information elements).
6.) The UDM is receiving the Registration Request message and is first using the Encrypted SUCI value for the look-up on which key is needed for the decryption. Now since the UDM knows which key is to be used and since it knows the UE, the UDM is taking the fetching the computed key KSUPI, is hashing this, and is fetching the encrypted SUCI from the local stored memory and is hashing the concatenated encrypted SUCI and the hashed key KSUPI. The outcome of this hashing (refers to the expected hash) will be compared with the received hash value.
7.) This refers to the registration and auth execution and completion. Rationale: The execution steps above refer basically to the auth of the encrypted SUCI and the authentication (legitimacy) of the UE. After this the normal Auth needs to be processed.
8.) After successful encrypted SUCI usage, both UE and UDM deletes this value from the list and same UE can’t use the same for further communications.
9.)10.)11.) This refers to the renewing and deployment of new list of encrypted SUCI values. UDM could use old RANDSUPI and continue to use the KSUPI for encryptions, but also UDM could decide to refresh this key KSUPI by creating new RANDSUPI and pass it to UE.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.13.3 Evaluation
| TBD
Editor’s Note: Further evaluation to be added.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.14 Solution #14 to SUCI Calculation: Symmetric solution on SUCI protection
| |
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.14.1 Introduction
| Editor’s Note: Analysis on the probability of desynchronization of eSUCIs is FFS.
Editor’s Note: Resynchronization of desynchronized eSUCIs is FFS.
Editor’s Note: The benefit of this solution is FFS.
The following are principles of the solution:
- UE is able to be provisioned with an enhanced SUCI (eSUCI) by UDM, or by pre-configuration, which is calculated with quantum resistant symmetric algorithm, symmetric home network key, and SUPI.
- If eSUCI is available, UE uses the eSUCI for initial Registration procedure.
- UDM calculates new eSUCI and updates towards UE after initial Registration procedure, UE does not calculate the eSUCI.
The following figure depicts the Encryption based on quantum resistant symmetric algorithm and symmetric key at the home network side.
Figure 7.2.1.14.1-1: Encryption based on symmetric key and algorithm at the Home Network
The following figure depicts the Decryption based on quantum resistant symmetric algorithm and symmetric key at the home network side.
Figure 7.2.1.14.1-2: Decryption based on symmetric key and algorithm at the Home Network
The Symmetric Key of Home Network, which is not a per UE key, is resident in SIDF/UDM and NOT shared with UE.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.14.2 Solution details
| The following figure depicts the initial Registration procedure using eSUCI, which is calculated with quantum resistant symmetric algorithm, symmetric home network key, and SUPI.
Figure 7.2.1.14.2-1: Initial Registration with eSUCI
1. If an enhanced SUCI (eSUCI), which is generated as described in clause 7.2.1.14.1, is provisioned during previous initial Registration procedure or pre-configured in UE (e.g., in NVM of ME or in USIM), the UE sends initial Registration Request (eSUCI) message to AMF/SEAF. If eSUCI is not available in UE, the UE uses asymmetric method (e.g. legacy or enhanced) to calculate a SUCI as an eSUCI for the initial Registration procedure.
Editor’s Note: Format of eSUCI is FFS.
Editor’s Note: Clarification on step 1 is ffs, e.g. proof-of-possession, exception case.
2. AMF/SEAF invokes Nausf_UEAuthentication_Authenticate Request (eSUCI) towards AUSF.
3. AUSF invokes Nudm_UEAuthentication_Get Request (eSUCI) towards SIDF/UDM.
4. SIDF decodes the eSUCI to get SUPI as described in clause 7.2.1.14.1 or using asymmetric method.
5. The UDM, AUSF, AMF/SEAF, and UE performs authentication procedure based on the SUPI decoded from the eSUCI.
6. If the authentication succeeds, SIDF/UDM calculates a new eSUCI as described in clause 7.2.1.14.1.
7. SIDF/UDM returns the calculated eSUCI to AUSF.
8. AUSF responds to AMF/SEAF with the new eSUCI.
9. AMF sends Registration Response (new eSUCI) to UE.
10. UE stores the new eSUCI, e.g. in the NVM of ME or in USIM, which will be used for successive initial Registration procedure.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.14.3 Evaluation
| TBD
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.15 Solution #15 to SUCI calculation: SUCI calculation with symmetric key
| |
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.15.1 Introduction
| This solution derives the encryption key EK, ICB and MAC key MK from the root key K. The encryption key length and MAC key length are increased to 256 Bit and AES-256-CTR is used for encrypting the SUPI.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.15.2 Solution details
| |
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.15.2.1 Processing on UE side
| The UE generates a 256 Bit Nonce#1 and similar to MILENAGE, the UE creates two additional Nonces by using a 64 Bit rotate operation. The Nonce#2 is created by rotating/shifting 64 Bits to the left of Nonce#1 and Nonce#3 by rotating/shifting 64 Bits to the left of Nonce#2.
The 256 Bit Encryption key EK is derived using a HMAC–SHA-256 with the root key K and the Nonce#1 as input.
The 128 Bit ICB is derived using a HMAC–SHA-256 with the root key K and the Nonce#2 as input with the output hash truncated to the 128 most significant bits.
The 256 Bit MAC key MK is derived using a HMAC–SHA-256 with the root key K and the Nonce#3 as input.
The EK and ICB are input to the AES-256-CTR, the output is a 256 Bit ciphertext of the encrypted SUPI.
The MK is used with the Nonce and the Ciphertext as input to a HMAC-SHA-256 function to generate a 256 Bit long MAC.
Since the computation is different, a 6G SUCI indication is required that the SUCI is differently concealed as in 5G.
The full SUCI has then the format as shown below with SUCI = 6G SUCI Indication || Nonce#1 || Ciphertext || MAC.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.15.2.2 Processing on home network side
| The home network detects the new SUCI format based on the 6G SUCI indication.
The home network creates based on the received Nonce#1 the two additional nonces Nonce#2 and Nonce#3.
The home network derives the 256 Bit Encryption key EK, 128 Bit ICB and 256 Bit MAC key MK in the same way as in the UE, using the Nonce#1, Nonce#2 and Nonce#3 respectively.
The home network verifies the MAC and decryptes the SUCI to SUPI.
Editor’s Note: For easier understanding of the solution described, further details on how to implement the solution (e.g., the schematic figures as in TS 33501 and call flows) is FFS.
Editor’s Notes: it is FFS how the network identifies which K to be used to derive the EK.
Editor’s Note: Details about the management of the root key, including generation, agreement, storage, revocation, etc. are FFS.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.15.3 Evaluation
| TBD
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.16 Solution #16 to SUCI calculation: Solution for PQC based SUCI Computation
| |
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.16.1 Introduction
| This solution address PQC algorithm based SUCI calculations.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.16.2 Solution details
| Processing on UE side:
Figure 1a: Encryption based on PQC shared key generation at UE
The UE computes a fresh SUCI, using the provisioned PQC-based public key of the home network (HN), and PQC-based key encapsulation mechanism (KEM) according to the parameters provisioned by home network as follows:
1. UE generates an ephemeral shared key and an encrypted PQC shared key based on a PQC-based public key associated with the home network. The PQC-based home network public key is identified using a HN PQC Public key ID or an existing HN Public key ID can indicate the HN PQC Public key with a related value.
2. UE generates ephemeral symmetric encryption key and ephemeral MAC key using a KDF function and ephemeral shared key along with input parameters such as Freshness parameter i.e., timestamp and SUCI Protection Profile ID.
3,4. UE protects the plaintext block (i.e. SUPI or UE ID), using the encryption key and the MAC key. The final output is the concatenation of Freshness Parameter, SUCI Protection Profile ID, encrypted PQC shared key and the ciphertext.
The final output i.e., scheme output coding is upto stage 3 similar to TS 23.003 [74]. The computed SUCI along with scheme output is sent from UE to network for authenticating the subscriber.
Processing on home network side
Figure 1b: Decryption based on PQC shared key generation at home network
For deconcealing the SUCI, the home network uses the received encrypted PQC shared key, and the PQC-based private key of the home network along with other parameters as described in the steps below:
1. Home network (HN) decapsulates the encrypted PQC shared key to derive the ephemeral shared key.
2. HN generates ephemeral symmetric (de)encryption key and ephemeral MAC key using a KDF function and derived ephemeral shared key along with input parameters such as such as Freshness parameter i.e., timestamp and SUCI Protection Profile ID.
3,4. HN verifies the MAC and decrypts the ciphertext to derive the plaintext block (i.e. SUPI or UE ID), using the MAC key and (de)encryption key respectively.
Example profile for SUCI Calculation: Profile C (PQC only): Profile C uses ML-KEM as defined in [21] to generate shared key Z1 integrated with AES encryption scheme.
The ME and SIDF implement this profile. The parameters for this profile are the following:
- ML KEM parameters : Level 3 (k, lattice dimension 3)
- KDF : ANSI-X9.63-KDF [9]
- Hash : SHA-256
- Shared secret key Z1 : Shared secret field from ML-KEM
- MAC : HMAC–SHA-256
- mackeylen : 32 octets (256 bits)
- maclen : 8 octets (64 bits)
- SharedInfo1 : N/A
- SharedInfo2 : the empty string
- ENC : AES–256 in CTR mode
- enckeylen : 32 octets (256 bits)
- icblen : 32 octets (256 bits)
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.1.16.3 Evaluation
| The solution has the following impacts:
New PQC algorithms and related profiles need to be supported by the UE and Network.
The UE generates ephemeral symmetric encryption key and ephemeral MAC key using a KDF function and ephemeral shared key along with input parameters such as Freshness parameter i.e., timestamp and SUCI Protection Profile ID. HN generates ephemeral symmetric (de)encryption key and ephemeral MAC key using a KDF function and derived ephemeral shared key along with input parameters such as such as Freshness parameter i.e., timestamp and SUCI Protection Profile ID. The use of timestamp and profile information as input allows replay protection for the SUCI and binds to the profile being used among multiple profiles respectively.
Editor’s Note 1: How the addition of freshness parameter is useful against an attack using CRQC is FFS.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.2 Solutions to MIKEY-SAKKE key exchange
| |
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.2.1 Solution #1 to MIKEY-SAKKE key exchange: mitigate
| |
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.2.1.1 Introduction
| There are a number of existing mitigations built into the Mission Critical system. Pending development of a post-quantum replacement for MIKEY-SAKKE it is possible these offer sufficient mitigation for threats, in particular harvest-now-decrypt-later. This is not proposed as a long term migration plan. That will require either a PQ identity based encryption scheme standard or a re-architecting the Mission Critical system.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 7.2.2.1.2 Solution Details
| In the on-network case, MIKEY-SAKKE key exchanges are protected by one or more layers of additional cryptographic protections as specified by clauses 5 and 6 in TS 33.180 [3]. Assuming these protocols, e.g. IPsec, are migrated to quantum-safe alternatives, this mitigates the risk of a passive attacker being able to harvest keys from the UE-to-MCX server interface.
There are further built-in protections for on-network access such as secure authentication to the network which further limit what an adversary can do with a forged signature on an I_MESSAGE.
Internal MCX interfaces over which I_MESSAGEs may be transferred may be protected by mTLS. This is currently optional but could be made mandatory.
It is also possible to re-use a security context established on-network when communicating off-network. Deployments could consider prohibiting off-network exchanges as one mitigation without further changes to the protocol.
All of the above is either within the standard already, or a configuration/policy for the UE. Further mitigations could be developed either as part of the standards or as informative text. There is a gap in the off-network case however this is currently out of scope.
Editor’s Note: Whether further mitigations for the off-network case can be considered is FFS.
7.2.2.1.3 Evaluation
Editor’s Note: This clause is FFS.
|
d20b4aa3005bbe8602412746b3658946 | 33.703 | 8 Conclusions
| Editor’s Note: This clause contains agreed conclusions and any normative work is recommended.
Annex A (informative):
Change history
Change history
Date
Meeting
TDoc
CR
Rev
Cat
Subject/Comment
New version
2025-08
SA3#123
S3-252632
TR 33.703 skeleton
0.0.0
2025-08
SA3#123
S3-252975
Incorporate pCRs from S3-252976, S3-252977, S3-252978, S3-252983, S3-252984, S3-253037
0.1.0
2025-10
SA3#124
S3-253686
Incorporate pCRs from S3-253687, S3-253688, S3-253689, S3-253691, S3-253692, S3-253847, S3-253693, S3-253694, S3-253695, S3-253696, S3-253830, S3-253831, S3-253486, S3-253832, S3-253833, S3-253855, S3-253835, S2-253836, S3-253837, S3-253838, S3-253839, S3-253841, S3-253840, S3-253842, S3-253843, S3-253844, S3-253845
0.2.0
2025-11
SA3#125
S3-254534
Incorporate pCRs from S3-254564, S3-254565, S3-254566, S3-254567, S3-254731, S3-254568, S3-254732, S3-254181, S3-254733, S3-254745, S3-254746, S3-254736, S3-254749, S3-254728
0.3.0
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 1 Scope
| The present document studies the security when a PLMN hosts an NPN with dedicated NFs deployed in the PNI-NPN operational domain, including:
1.Key issues and potential security requirements for the scenario of PLMN hosting a NPN where the interfaces between PLMN operational domain and PNI-NPN domain include N9. And solutions to address the identified security requirements.
2.Evaluation of the security recommendations given in TS 33.501[2] annex AB apply to the scenario of PLMN hosting a NPN where more CP functions (except AMF, SMF, UDM) are deployed in PNI-NPN domain.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 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 TS 33.501: "Security architecture and procedures for 5G system"
[3] 3GPP TR 33.757: "Study on security for a PLMN hosting a Non-Public Network (NPN)"
[4] 3GPP TS 23.501: " System architecture for the 5G System (5GS); Stage 2"
[5] 3GPP TS 29.281: "General Packet Radio System (GPRS) Tunnelling Protocol User Plane (GTPv1-U)".
[6] Yiming Zhang, et al. “Invade the Walled Garden: Evaluating GTP Security in Cellular Networks”, IEEE Symposium on Security and Privacy (SP), May 2025.
…
[x] <doctype> <#>[ ([up to and including]{yyyy[-mm]|V<a[.b[.c]]>}[onwards])]: "<Title>".
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 3 Definitions of terms, symbols and abbreviations
| |
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 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.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 3.2 Symbols
| For the purposes of the present document, the following symbols apply:
<symbol> <Explanation>
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 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>
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 4 Architecture
| TR 33.757[3] has studied two scenarios of PLMN hosting a NPN, where the interface between PLMN operational domain and PNI-NPN domain is N4 or SBA interface.
Figure 4-1 N9 interface across PLMN operational domain and PNI-NPN operational domain
In addition to the scenarios in TR 33.757[3], the interfaces between PLMN operational domain and PNI-NPN domain can include N9. Considering the scenario depicted in Figure 4-1, the dedicated UPF in PNI-NPN operational domain2 is controlled by SMF in service area B, and customers can access the DN through the UPF in service area A or the UPF in service area B depending on customers’ location. The situation is similar for the dedicated UPF in PNI-NPN operational domain1.
Editor’s Note: More clarification on the architecture is FFS.
In TR 33.757[3], the CP functions deployed in the PNI-NPN operational domain only consider AMF and SMF. However, more CP functions (except AMF, SMF, UDM) defined in TS 23.501 [4] are likely to be deployed in the PNI-NPN operational domain.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 5 Security assumptions
| Editor’s Note: This clause includes the security assumptions for the study.
The security assumption in TR 33.757[3] clause 5 apply.
Editor’s Note: Further security assumption is FFS.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 6 Evaluation for SBA interface protection
| Editor’s Note: This clause evaluate if security recommendations given in TS 33.501[2] annex AB apply to the scenario of PLMN hosting a NPN where more CP functions (except AMF, SMF, UDM) are deployed in PNI-NPN domain.
The 5G System architecture consists of the network functions is list in TS 23.501[4] clause 4.2.2, while the service-based interface is list in TS 23.501[4] clause 4.2.6.
The following NFs specified in TS 23.501[4] clause 4.2.2 with service-based interface specified in TS 23.501[4] clause 4.2.6 may be considered not to be deployed in the PNI-NPN operator domain:
- Authentication Server Function (AUSF).
- Unified Data Management (UDM).
- Unified Data Repository (UDR).
- Unstructured Data Storage Function (UDSF).
- 5G-Equipment Identity Register (5G-EIR).
- CHarging Function (CHF).
Except the NFs list above, the NFs specified in TS 23.501[4] clause 4.2.2 with service-based interface specified in TS 23.501[4] clause 4.2.6 may be considered to be deployed in the PNI-NPN operator domain.
The security recommendations given in TS 33.501[2] annex AB apply to the NF which is considered to be deployed in the PNI-NPN operator domain.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 7 Key issues
| Editor’s Note: This clause contains all the key issues identified during the study.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 7.1 Key Issue #1: TEID issue in N9 interface
| |
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 7.1.1 Key issue details
| A UPF can be deployed in the PNI-NPN operational domain and connects to a UPF deployed in the PLMN operational domain via N9 interface. Attackers in PNI-NPN operational domain (e.g., a misbehaving employee in PNI-NPN or an external attacker gaining unauthorized access to the PNI-NPN networks) can obtain the TEID from the UPF deployed in PNI-NPN operational domain.
For example, TS 29.281[5] clause 5.1 states:
Tunnel Endpoint Identifier (TEID): This field unambiguously identifies a tunnel endpoint in the receiving GTP‑U protocol entity. The receiving end side of a GTP tunnel locally assigns the TEID value the transmitting side has to use. The TEID value shall be assigned in a non-predictable manner......
UPFs can select the first TEID in a non-predictable manner (e.g., randomly) but allocate subsequent TEID numbers sequentially.
Furthermore, TS 29.281[5] clause 7.3.1 states:
When a GTP-U node receives a G-PDU for which no EPS Bearer context, PDP context, PDU Session, MBMS Bearer context, or RAB exists, the GTP-U node shall discard the G-PDU. If the TEID of the incoming G-PDU is different from the value 'all zeros' the GTP-U node shall also return a GTP error indication to the originating node.
As a TEID without an established context will trigger error codes in the response while a correct TEID will not, allowing an attacker to guess whether a TEID is used effectively.
Figure 7.1-1 Scenario involving N9 interface and having TEID issue
After an attacker in PNI-NPN operational domain1 obtain the TEID assigned by the PLMN UPF to UPF in PNI-NPN operational domain1, the attack can use this information to infer the TEIDs assigned by the PLMN UPF to UPF in PNI-NPN operational domain2, PLMN gNBs, SMF(through N4-u). The attack can further use the TEIDs to hijack subscriber traffic in other GTP tunnels, as described in the research paper "Invade the Walled Garden: Evaluating GTP Security in Cellular Networks"[6]. More specifically, as illustrated in Figure 7.1-1, the attacker in PNI-NPN operational domain1 can perform the following attacks:
- Attack to other PNI-NPN: The attacker sends a GTP-U PDU message to UPF3 that contain TEID2 (corresponding to the legitimate UPF2→UPF3 GTP-U tunnel)—with the inner packet whose destination IP address is that of a UE which is allowed to access PNI-NPN operational domain2 from PLMN. Since the message matches the PDR corresponding to the legitimate UPF2→UPF3 GTP-U tunnel, UPF3 will forward the messages to the UE according to the related FAR. Similarly, the attacker can send a GTP-U PDU message to UPF3 that contain TEID3 with the inner packet whose source IP address is that of a UE which is allowed to access PNI-NPN operational domain2 from PLMN. UPF3 will forward the messages to UPF2 according to the related FAR. In this way, an attacker in PNI-NPN operational domain1 can send malicious messages to attack UEs which are allowed to access PNI-NPN operational domain2 from PLMN, and also target UPF2 and DN2.
- IP address fraud: The attacker sends a GTP-U PDU message to UPF3 that contain TEID4(corresponding to the legitimate SMF→UPF3 N4-U tunnel)—with the inner packet carrying spoofed IPv6 RA. UPF3 will forward the messages to the UE according to the related FAR. This can cause the UE to adopt the spoofed IPv6 address prefix, ultimately disrupting its connection with the 5GC.
- Bill inflation: The attacker sends a GTP-U PDU message to UPF3 that contain TEID2(corresponding to the legitimate UPF2→UPF3 GTP-U tunnel)—with the inner packet whose source IP address is that of a UE. In this way, the attacker can inflate the victim’s bill by (silently) sending large amounts of traffic.
The KI aims to evaluate whether the requirement on TEID unpredictability in TS 29.281[5] is enough for the case of N9 interface, and whether improved/refined requirements are needed for N9 interface. The KI does not aim to define the format of TEID.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 7.2.2 Security threats
| When there is no security enabled on the N9 interface between PLMN operation domain and PNI-NPN operation domain, attackers in the PNI-NPN or PLMN operational domain can launch attacks to PLMN or NPN over the intersection.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 7.3.3 Potential security requirements
| TBD.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 7.2 Key Issue #2: Inter domain security on N9 interface
| |
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 7.2.1 Key issue details
|
Figure 7.2-1 Scenario involving N9 interface
Considering the scenario depicted in Figure 7.2-1, attackers in PNI-NPN or PLMN operational domain (e.g., a misbehaving employee in PNI-NPN, PLMN or an external attacker gaining unauthorized access to the PNI-NPN or PLMN networks) can attack the opposing domain through the N9 interface.
TR 33.757[3] studied the intersection between the SMF and UPF and potential solution which could be used to improve resilience at the intersection. This KI proposes to improve the resilience of the N9 interface end points, when used to communicate over the intersection, without injecting new functions in the intersection nor change GTP protocol. As the N9 interface is key, in the home routed roaming architecture, improvements have already been standardized for the inter-PLNM which do not apply of the case of PLMN and NPN interconnection.
The KI aims to evaluate whether existing security improvements for home routed roaming can be reused for the case of PLNM interacting with an NPN and vice versa.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 7.2.3 Potential security requirements
| The 5G system shall support a mechanism to protect the endpoints of the N9 interface between PLMN operation domain and PNI-NPN operation domain.
7.X Key Issue #X: <Key Issue Name>
7.X.1 Key issue details
7.X.2 Security threats
7.X.3 Potential security requirements
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 8 Solutions
| Editor’s Note: This clause contains the proposed solutions addressing the identified key issues.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 8.1 Mapping of solutions to key issues
| Editor's Note: This clause contains a table mapping between key issues and solutions.
Table 7.1-1: Mapping of solutions to key issues
Solutions
KI#X
KI#Y
KI#Z
8.Y Solution #Y: <Solution Name>
8.Y.1 Introduction
Editor’s Note: Each solution should list the key issues being addressed.
8.Y.2 Solution details
8.Y.3 Evaluation
Editor’s Note: Each solution should motivate how the potential security requirements of the key issues being addressed are fulfilled.
|
873ee27c242a05b1ac21717ead7f30a2 | 33.758 | 9 Conclusions
| Editor’s Note: This clause contains the agreed conclusions that will form the basis for any normative work.
Annex <X>:
Change history
Change history
Date
Meeting
TDoc
CR
Rev
Cat
Subject/Comment
New version
2025-10
SA3#124
S3-253336
Skeleton
0.0.0
2025-10
SA3#124
S3-253726
S3-253365, S3-253737, S3-253848, S3-253739, S3-253740 implemented
0.1.0
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 1 Scope
| The present document investigates and identifies the security threats, requirements and potential solution for Integrated Sensing and Communication (ISAC). Based on the architecture and system level enhancements studied in TR 23.700-14 [2], the work in this document focuses on the security and privacy aspects of gNB-based sensing for aerial object (e.g. drone) sensing target use cases.
The aerial object sensing target uses cases defined by TS 22.137 [3] and TR 22.837 [4] serve either the purpose of public safety, or as requested by the management entity (UAV management department, USS or UTM), without the necessity to identify the object.
Specifically, the present document covers the following:
- The identified key issues, threats, potential requirements and solutions for security protection during the service operations and procedures supporting Sensing services;
- The identified key issues, threats, potential requirements and solutions for protecting privacy for sensing data collection, sensing data processing, and sensing data exposure.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 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 23.700-14: "Study on Integrated Sensing and Communication; Stage 2".
[3] 3GPP TS 22.137: "Service requirements for Integrated Sensing and Communication; Stage 1".
[4] 3GPP TR 22.837: "Feasibility Study on Integrated Sensing and Communication".
[5] 3GPP TR 33.501: "Security architecture and procedures for 5G system".
[6] 3GPP TS 33.310: "Network Domain Security (NDS); Authentication Framework (AF)".
[7] 3GPP TS 33.210: "3G security; Network Domain Security (NDS); IP network layer security”.
[8] IETF RFC 6749: "The OAuth 2.0 Authorization Framework".
[9] 3GPP TS 23.501: "System Architecture for the 5G System".
[10] 3GPP TS 33.122: "Security Aspects of Common API Framework for 3GPP Northbound APIs".
[11] IETF RFC 6083: "Datagram Transport Layer Security (DTLS) for Stream Control Transmission Protocol (SCTP)".
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 3 Definitions of terms, symbols and abbreviations
| |
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 3.1 Terms
| For the purposes of the present document, the terms given in 3GPP 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 3GPP TR 21.905 [1].
example: text used to clarify abstract rules by applying them literally.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 3.2 Symbols
| For the purposes of the present document, the following symbols apply:
<symbol> <Explanation>
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 3.3 Abbreviations
| For the purposes of the present document, the abbreviations given in 3GPP 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 3GPP TR 21.905 [1].
<ABBREVIATION> <Expansion>
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 4 Architecture and security assumptions
| The following architecture and security assumptions are applied to the study:
- The architecture assumptions and principles for Integrated Sensing and Communication as defined in TR 23.700-14 [2] are used as architecture assumptions in this study.
- The security architecture, procedures, and security requirements for 5GS as defined in TS 33.501 [5] are used as a baseline.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5 Key issues
| Editor's Note: This clause contains all the key issues identified during the study.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.1 Key Issue #1: Security of sensing service authorization and sensing result exposure
| |
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.1.1 Key issue details
| In TR 23.700-14 [2], architecture for sensing services is studied to enable the 3GPP network to support sensing service invocation and revocation from the service consumer, and sensing result exposure to the service consumer.
Solutions addressing the KI#2 in TR 23.700-14 [2] of authorization and revocation for particular sensing services are developed, which focus on service request authorization or revocation based on the information of the service level agreement. Security aspects need to be discussed for the above mentioned procedures.
NOTE: Security aspects of sensing service revocation triggered by sensing service consumer is addressed in this key issue.
In addition, KI#5 in TR 23.700-14 [2] addresses the type of sensing result to be exposed and the method for the network to expose the sensing result to the service consumer. Security aspect of the exposure procedure also needs to be investigated.
This key issue is related to KI#2 and KI#5 of TR 23.700-14 [2] and addresses the security aspects for sensing service invocation, revocation, and sensing result exposure procedures between the network and sensing service consumer.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.1.2 Security threats
| Without proper authentication and authorization for sensing service, unauthorized party may be able to access to sensing service.
If the connection between sensing service consumer and NEF/SF is not protected, the attacker can tamper, inject, sniff or replay messages related to sensing service invocation, revocation and sensing result exposure.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.1.3 Potential security requirements
| The 5G system shall be able to support mutual authentication between sensing service consumer and NEF/SF.
The 5G system shall be able to support integrity protection, confidentiality protection and replay protection for the communication between sensing service consumer and NEF/SF.
The 5G system shall be able to authorize sensing service request from a sensing service consumer.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.2 Key Issue #2: Security protection for sensing service operations
| |
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.2.1 Key issue details
| According to TR 23.700-14 [2], after the sensing service request from the service consumer is authorized by the network, sensing service operations will be triggered and performed by the relevant network functions, which communicate with each other to obtain the sensing result.
In TR 23.700-14 [2], there are multiple solutions proposing sensing service operation procedures supported by sensing entities and different sensing related network functions (e.g. NEF, SF). The NEF needs to discover and select the SF to trigger sensing service operation. The SF needs to select proper sensing entity to collect sensing data in a specific sensing mode. When any of the service conditions of a sensing service is no longer met, an ongoing sensing service can be revoked by the network. The security aspects of all these sensing operations and procedures are to be addressed in this key issue.
NOTE 1: Security aspects of service operation revocation triggered by sensing function is addressed in this key issue, as it can be viewed as one type of sensing service operations.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.2.2 Security threats
| As the sensing service operations are performed among sensing function(s) and sensing entities, if the 5GC does not support sensing service operation authorization, the sensing service operation can be abused.
If the connection between sensing entity and sensing function is not securely established, an attacker is able to tamper or inject or replay sensing control messages and sensing data, or sniff the collected sensing data.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.2.3 Potential security requirements
| The 5G system shall be able to support authorization for sensing service operations.
The 5G system shall be able to support integrity protection, confidentiality protection and replay protection for the connection between sensing entity and SF.
Editor’s Note: More security requirements will be added depends on SA2 progress.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.3 Key issue #3 on privacy for sensing
| |
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.3.1 Key issue details
| This key issue focuses on the privacy aspect of sensing.
The introduction of sensing capabilities enables the network to collect and process sensing data about objects in the public or even private environment and expose derived sensing results, all without the direct participation or awareness of the sensed object. Considering that the sensing data or sensing result can contain privacy sensitive information, the privacy aspect of sensing service needs to be investigated.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.3.2 Security threats
| If any privacy related information is contained in the sensing data and is leaked to an unauthorized party, it could lead to privacy violation.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.3.3 Potential security requirements
| The 5G system shall provide a mechanism to mitigate privacy threats in the sensing system.
Editor's Note: further refinement of the above requirement is FFS.
Editor’s Note: whether this key issue needs 3GPP solution(s) is FFS, as there may be mechanism out-of-3GPP.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.4 Key issue #4 on active attacks in sensing
| |
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.4.1 Key issue details
| One of the use cases for sensing technology is detection of aerial objects. In order to be useful, the result has to be reliable, ie. report an aerial object when there is one, and report empty airspace only when the airspace is in fact empty.
During the sensing operation, an attacker could generate a radio signal that would confuse the receiving sensing node into determining that there is a aerial objects at a location where there is none (e.g by sending a signal that exhibits the typical micro-Doppler shift typical for UAV rotors), or into determining that there is no aerial object where in fact there is one (e.g. by generating noise such that the response by a real aerial object is drowned out, or perceived to come from a different location).
Editor's note: feasibility of the attack is FFS
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.4.2 Security threats
| Editor's note: threat description is FFS
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.4.3 Potential security requirements
| Editor's note: Requirements are FFS
Editor's note: Whether or not to coordinate with RAN1 is FFS
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.5 Key issue #5 on unauthorized passive sensing
| |
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.5.1 Key issue details
| The sensing mode considered in the present document is a collocated sensing transmitter and receiver. However, the sensing signal sent by the sensing transmitter is not only reflected to the collocated sensing receiver, but also attenuated and scattered in all directions. Therefore, it can be possible for an attacker to set up a sensing receiver that is not collocated with the sensing transmitter, thus allowing the attacker to perform it's own sensing. The difference with an attacker performing monostatic sensing on its own is that the attacker doesn't need to become active, thus minimizing the risk of being detected. In addition, the sensing transmitter of the operator's sensing infrastructure may be in a better location, i.e. higher up or closer to the target object.
Editor's note: Feasibility of the attack is FFS
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.5.2 Security threats
| Editor's note: threat description is FFS
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 5.5.3 Potential security requirements
| Editor's note: requirements are FFS.
Editor's note: Whether or not to coordinate with RAN1 is FFS.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6 Solutions
| Editor's Note: This clause contains the proposed solutions addressing the identified key issues.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.0 Mapping of solutions to key issues
| Editor's Note: This clause contains a table mapping between key issues and solutions.
Table 6.1-1: Mapping of solutions to key issues
Solutions
KI#1
KI#2
KI#3
KI#4
KI#5
#1.1
X
#1.2
X
#1.3
X
#1.4
X
#1.5
X
#1.6
X
#1.7
X
#2.1
X
#2.2
X
#2.3
X
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1 Solutions to KI#1
| |
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.1 Solution #1.1: Authorization for sensing service request from AF
| |
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.1.1 Introduction
| This solution addresses Key Issue #1: Security of authorization for sensing service invocation and revocation.
In this solution, the sensing service consumer is assumed to be an external AF. The NEF performs the access authorization by verifying the AF's identity, and the SF performs the service authorization by validating the feasibility and policy compliance of the specific sensing request parameters against network capabilities and operator rules.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.1.2 Solution details
| This solution proposes mutual certificate-based authentication between the NEF and the external AF/sensing service consumer using TLS. Certificate based authentication follows the profiles given in 3GPP TS 33.310 [6], clause 6.1.3a. The identities in the end entity certificates is used for authentication and policy checks.
For the protection of communication between AF/sensing service consumer and NEF, TLS is used to provide integrity protection, replay protection and confidentiality protection for the interface between the NEF and the AF/sensing service consumer. Security profiles for TLS implementation and usage follow the provisions given in clause 6.2 of TS 33.210 [7].
After the authentication, the following procedures are used for authorizing sensing service request.
Figure 6.1.1.2-1: Procedure for sensing service authorization
1. The AF sends sensing service request message to the NEF. The message includes AF ID, OAuth 2.0 token, and sensing service related parameters (e.g., target sensing area, sensing time, sensing type, etc).
NOTE 1: Details of the sensing authorization information for sensing service are out of scope of this solution.
2. NEF performs the authorization check for the sensing service request. This includes:
- validating the OAuth 2.0 token presented by the AF; and
- checking the AF's subscription profile to verify that the AF is entitled to request the sensing service.
If the check fails, the NEF rejects the request with a failure cause.
3. If the AF is authorized by the NEF to request for sensing service, the NEF discovers and selects the SF, and sends the Sensing service request message to the SF. This message includes sensing service related parameters.
4. The SF performs sensing service authorization based on the sensing service related parameters. Specifically, this includes:
- validating the sensing service related parameters against operator-defined service policies (e.g., restricted zones, restricted time); and
- checking if the network has available resources to fulfill the request.
If the authorization fails, the SF rejects the request with a failure cause. The reject message is sent to AF via NEF.
NOTE 2: Validations performed by SF for sensing service authorization are out of scope of this solution.
5. After successful authorization, the SF proceeds to execute the sensing service.
6-7. The SF provides sensing results in sensing service response to the AF via NEF.
NOTE 3: Details of the service procedures are out of scope of this solution.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.1.3 Evaluation
| This solution is based on the assumption that the sensing service consumer is an external AF. This solution does not address authorization for internal AF.
This solution proposes to reuse the existing mechanism to perform mutual authentication and secure communication between sensing service consumer and NEF. Details of the OAuth framework is not addressed in this solution.
The following impacts are needed:
- NEF needs to perform the access authorization by verifying the AF's identity;
- SF needs to perform sensing service authorization.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.2 Solution #1.2: Authorization for Sensing Service
| |
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.2.1 Introduction
| This solution addresses requirements of key issue #1.
In this solution, existing SBA security framework is reused so that authentication and communication protection among sensing service consumer and NEF/SF can be protected using existing SBA mechanism, for authorization, NRF is deemed as authorization check point, and some specific criteria for sensing, e.g. sensing service, sensing location, is considered for sensing service authorization.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.2.2 Solution details
|
Figure 6.1.2.2 - Authorization for Sensing Service
1. AF sends sensing service request to the NEF. The message includes the AF ID, the requested sensing services and optionally the requested sensing location for the sensing service.
2. The NEF performs SF discovery procedure via NRF.
3. The NEF sends token request the NRF as described in clause 7.1.3 of TR 23.700-14 [2].
4. The NRF performs service authorization, the NRF checks whether the AF is authorized to access the SF, whether the requested sensing services are allowed for the AF, and optionally whether the requested sensing location for the sensing service is allowed.
5. If it is authorized, NRF will issue access token and send it to the NEF, and the access token claim includes AF ID, the allowed sensing services and optionally the allowed sensing location.
6. The NEF sends sensing service request to the SF. The message includes the access token, the AF ID, the requested sensing services and optionally the requested sensing location for the sensing service.
7. After successfully verifying the access token, the SF performs sensing based on the allowed sensing services and optionally the allowed sensing location. The SF sends sensing service response including the sensing result to the NEF.
8. The NEF sends sensing service response including the sensing result to the AF.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.2.3 Evaluation
| The solution address authorization requirements of key issue #1.
The solution reuses existing SBA framework for token-based authorization.
According to conclusion made for static authorization in table 7.1.2-1 of TS 23.700-14 [2], AF ID, sensing area for sensing, sensing service type are criteria for authorization for sensing service, the solution proposes the similar principles for token-based authorization, those criteria will be checked by NRF and reflect on issued token.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.3 Solution #1.3: Solution on authorization for sensing service request
| 6.1.3.1 Introduction
This solution addresses Key Issue#1 on Security of authorization for sensing service invocation and revocation. Specifically, it addresses the third requirement in KI#1: “The 5G system shall be able to authorize sensing service request from a sensing service consumer”.
According to TR 23.700-14 [2], a sensing service request may be initiated by a sensing service consumer. The authorization on service permission includes two levels:
- The first level of authorization is for service access. When the NEF receives the sensing service request initiated by the sensing service consumer (e.g. an AF), the NEF can determine whether the sensing service consumer is authorized to request the sensing service from the 5GC, according to clause 12 in TS 33.501 [5].
- The second level of authorization is based on the local policy. The Sensing Function may check the Sensing Profile to verify the sensing service request from NEF to determine if a sensing service is allowed.
6.1.3.2 Solution details
Figure 6.1.3.2-1: Authorization for sensing service request
1. The AF requests a service request for sensing. The request may include AF ID, sensing service type (object detection, object tracking, etc), sensing service requirements (e.g. accuracy, latency, etc), sensing service area.
2. The NEF may authorize the sensing service request from the AF by reusing the OAuth 2.0 mechanism in clause 12 of TS 33.501 [5].
3. The NEF may discover and select the candidate Sensing Function(s).
4. If the authorization succeeds, then the NEF sends the sensing service request message to the Sensing Function. The request message may contain AF ID, sensing service area, sensing service type, sensing service requirements.
5. The Sensing Function may authorize the sensing service request based on the local policy. The Sensing Function may check the Sensing Profiles to verify the sensing service request from NEF, which may contain allowed sensing service area, allowed sensing service type, allowed sensing service time duration, etc.
NOTE 1: The Sensing Profile is stored in Sensing authorization functionality for authorisation of the sensing service request, e.g., Sensing Function.
NOTE 2: The details of Sensing Profiles are out of scope of this solution.
6. If the authorization succeeds, then the Sensing Function proceeds to execute the sensing service.
7. The Sensing Function sends the sensing results to NEF.
8. The NEF sends the sensing results to AF.
6.1.3.3 Evaluation
This solution addresses the KI#1: “The 5G system shall be able to authorize sensing service request from a sensing service consumer”.
This solution reuses the OAuth 2.0 mechanism defined in TS 33.501 [5] to address the authorization of AF for sensing service request in NEF.
In this solution, the authorization of AF's Sensing Service Request in Sensing Function is aligned to TS 23.700-14 [2].
No new security mechanism is introduced.
Editor’s Note: Whether the solution fulfills all SA2 use cases is FFS.
6.1.4 Solution #1.4: Security of the connection to the Sensing service consumer
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.4.1 Introduction
| This solution aims to address the security requirements in Key Issue #1. In TR 23.700-14 [2], architecture for sensing services is studied to enable the 3GPP network to support sensing service invocation and revocation from the service consumer.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.4.2 Solution details
| The Sensing service consumer acts as external Application Function (AF) to interact with the network.
If the Sensing service consumer acting as external AF then it only interacts with network via NEF. In this case the security mechanisms in clauses 12 of [5] are reused to provide mutual authentication, authorisation, integrity protection, confidentiality protection and replay protection between Sensing service consumer and the NEF.
Editor’s Note: the architecture needs to inline to SA2
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.4.3 Evaluation
| TBD.
6.1.5 Solution #1.5: authorize sensing service request using OAuth-based authorization mechanism
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.5.1 Introduction
| The solution addresses KI#1 to authorize sensing service request from the sensing service consumer
Key issues related to System Architecture to Support Sensing, Authorization and Revocation to Support Sensing Service, and Sensing Result Exposure are studied in TR 23.700-14 [2]. Based on solutions for those KIs, a sensing service consumer may access sensing service from sensing function indirectly via NEF. For example, if the sensing service consumer is external AF, it accesses the sensing function through NEF. The sensing service request may trigger operation or revocation of sensing on specific object in specific area at specific accuracy level during specific time, or subscribe to specific sensing result. Sensing service authorization polices are defined in some solutions, and local policies-based authorization is also discussed in some solutions.
If the sensing service consumer is external AF, as specified in clause 12 of TS 33.501 [5], the NEF shall authorize the requests from AF using OAuth-based authorization mechanism, the specific authorization mechanisms shall follow the provisions given in RFC 6749 [8]. When the NEF supports CAPIF for external exposure as specified in clause 6.2.5.1 in TS 23.501[9], then CAPIF core function shall choose the appropriate CAPIF-2e security method as defined in the sub-clause 6.5.2 in TS 33.122[10] for mutual authentication and protection of the NEF – AF interface.
If the sensing service consumer is an AF inside the operator’s domain, according to clause 13 of TS 33.501 [5] and clause 6.2.10 of TS 23.501, OAuth 2.0 based authorization is reused. Static authorization is based on local authorization policy at the SF and can be used when token-based authorization is not used.
In general, OAuth 2.0 based authorization can be reused to authorize sensing service request from sensing service consumer.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.5.2 Solution details
| 6.1.5.2.1 Sensing service consumer is an AF inside the trusted domain
Precondition:
• OAM provisions sensing authorization policies in NRF enabling which sensing consumers are allowed to access / trigger what type of sensing operation on which kind of object in which area at which time with what level of accuracy.
• SF registers to NRF with NF profile including supporting sensing objects, sensing area, sensing accuracy, etc.
• AF registers to NRF with profile including NF/AF Id, type, location, etc.
1. AF sends request to NRF to discover potential sensing functions for the required sensing service.
2. AF sends Access token request for Sensing Service (e.g. sensing service type, sensing service area, sensing duration, sensing quality of service requirements)
3. NRF Authorizes the request based on the required sensing service, area, duration, accuracy, sensing consumer profile and preconfigured policies, etc.
4. NRF sends Access Token response including sensing related claims
5. AF sends sensing service request to a discovered SF with access token got in step 4
6-8. SF validates the token, triggers sensing operation and sends response to the AF.
6.1.5.2.2 Sensing service consumer is external AF
Precondition:
• OAM may provision operation access control policies and sensing authorization policies in NEF, enabling which AFs are allowed to access what type of sensing operation on which kind of object in which area at which time with what level of accuracy.
• SF registers to NRF with NF profile including supporting sensing objects, sensing area, sensing accuracy, etc.
1. SSC sends Sensing Service Request (e.g. sensing service type, sensing service area, sensing duration, sensing quality of service requirements), with the access token obtained from the Authorization Server.
NOTE: if CAPIF is supported, the token will be generated by CCF, which will be aware of the sensing policies.
2-4. If the information in the access token is not sufficient to authorize the request, NEF retrieves detail sensing authorization polices from other sensing specific policy function, and retrieve sensing services registered to NRF, and authorizes the request based on the required sensing service against available sensing services got from NRF, sensing consumer information, operation access control policies and sensing authorization policies.
5-6. NEF discovers a SF based on the request and forwards the sensing service request to the SF, and sends response to the SSC.
NOTE: NEF follows existing discovery and authorization procedure defined in TS 23.502 and 33.501 [5] to discover and access the SF.
7. SF triggers sensing operation.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.5.3 Evaluation
| Editor’s Note: Each solution should motivate how the potential security requirements of the key issues being addressed are fulfilled.
6.1.6 Solution #1.6: Sensing Service Authorization at the Sensing Function
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.6.1 Introduction
| This solution addresses the potential authorization requirement of Key Issue #1: Security of authorization for sensing service invocation and revocation:
“The 5G system shall be able to authorize sensing service request from a sensing service consumer..”
It is proposed that the Sensing Function performs the authorization of the Sensing Request.
|
03f08e095da0e7d00cf764af0a02ab5a | 33.777 | 6.1.6.2 Solution details
|
Figure 6.1.6.2-1: Sensing Service Authorization at the Sensing Function
1. It is assumed the Sensing Service Consumer (AF) and the NEF have a security association as described in 3GPP TS 33.501 [5], clause 12 “Security aspects of Network Exposure Function (NEF)”. The Sensing Service Consumer sends a Sensing Service Request to the NEF with the descriptive information e.g. sensing service type (object detection, object tracking, environment sensing, etc.), sensing service requirements (e.g. accuracy, latency, resolution, etc.) and time information when the sensing service is needed (e.g. time for sensing measurement, time for sensing report) etc.
2. The NEF selects a Sensing Function for invoking the Sensing Service and to authorize the request. The NEF sends a Nsf_Sensing _Authorization_Request including the AF ID and the sensing information received from the Sensing Service Consumer to the Sensing Function.
3. The Sensing Function fetches Sensing Profile Information for the AF ID. The Sensing Function performs the authorization of the sensing request from the NEF by verifying whether the information from the Sensing Request matches the information stored in the Sensing Sensing Authorization information for the AF.
Table 6.1.6.2-1: Sensing Authorization information for Sensing Service [2]
AF Authorization Data
Description
AF ID
Identifier used to identify the AF.
Allowed/Not allowed area for sensing
Indicate the allowed/not allowed area for the indicated AF to trigger the sensing services operations (NOTE 1).
Allowed/Not allowed time period for sensing
Indicate the allowed time period within which AF can trigger a particular sensing service operation.
(Allowed) sensing service type
Indicate the allowed sensing service type (e.g. object tracking, detection) for the indicated AF to trigger the sensing services operations.
NOTE 1: The Allowed area and the Not allowed area may be both present or only one of them can be present, for example if the AF is allowed to make a request for the majority of PLMN coverage excluding the not allowed area only the not allowed area can be present. If both are not present all area is allowed.
If the sensing service authorization is successful, the Sensing Function initiates the sensing procedure with the corresponding NF. If the sensing service authorization fails, the Sensing Function responds the failure to the NEF.
4. The Sensing Function responds to the NEF either with the sensing information result from the sensing procedure, or with a successful authorization response or with an authorization failure response.
Editor’s Note: The messages in step 2 and step 5 need to be aligned with SA2.
5. The NEF forwards the message from the Sensing Function to the AF.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.