Reviewing the role of machine learning and artificial intelligence for remote attestation in 5G+ networks

Abstract

The next generation of communication networks promises an exponential growth in number, low latency, and heterogeneity of devices. Consequently, we need to be able to trust devices and device-to-device interactions in sections of a 5G+ network, commonly known as slices. Devices must be willing and able to remotely attest to their trustworthiness. Although trust has previously been based upon deterministic and hardware-driven protocols, over recent decades it has become more common to incorporate artificial intelligence and machine learning (AI/ML) modeling to supplement those protocols. In this paper we review some key aspects of models used for trust of devices, including important criteria for model selection, model structure and inputs, and advantages and disadvantages of these models. We also examine how these AI/ML models intersect with 5G network architecture. Following that, we discuss what sort of data are expected for these trust models. Finally, we discuss next steps for AI/ML models for remote attestation in 5G+ networks.