Feasibility and Security of AAG-Type Key Exchange Protocols in 5G Networks

Abstract

Fifth-generation (5G) mobile networks rely on a layered security architecture in which symmetric cryptography ensures low-latency data protection, while asymmetric mechanisms support identity privacy and key establishment. Current deployments primarily depend on classical Diffie-Hellman (DH) schemes based on commutative key exchange. However, the long term hardness of the discrete logarithm problem, together with the emergence of post-quantum threats, motivates the exploration of alternative algebraic foundations for key exchange. In this work, we investigate whether non-commutative key exchange protocols, specifically the Anshel-Anshel-Goldfeld (Anshel-Anshel-Goldfeld (AAG)) scheme, can reinforce Diffie-Hellman (DH) within Fifth Generation (5G) systems. We study both DH and AAG, evaluating computational overhead, shared key derivation latency, and end-to-end delay across multiple hardware platforms representative of the User Equipment (UE) and Home Network (HN) roles in 5G networks. Furthermore, we examine search-space size and structural attack surfaces to provide a comprehensive assessment of security. Our results show that, although AAG entails higher computational overhead compared to DH, its integration remains feasible within the latency limits defined by 3GPP, particularly on network-side hardware. Additionally, the AAG exchange benefits from a substantially larger private key space for comparable parameter sizes. Consequently, the cost of an attack is significantly greater for AAG, while the computational burden on legitimate participants remains minimal.