A Hybrid Reinforcement Learning-Based Trust Model for 5G Networks

Abstract

Trust investigation in 5G, which is the next-generation wireless network, is still at its infancy. This research proposes a hybrid trust model for the selection of a legitimate (or trusted) forwarding (relay) entity and to countermeasure intelligent attacks against a route selection scheme. The hybrid trust model has a centralized entity (i.e., a centralized controller C <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">c</inf> ) that provides the security level of the operating environment to network entities, and distributed entities that identify malicious or legitimate entities in the network. When the security level of the operating environment is high (low), the distributed entities learn more (lesser) from their respective operating environment. While legitimate entities can use artificial intelligence, such as reinforcement learning (RL), to enhance their trust models, the malicious entities can also use artificial intelligence to increase the detrimental effects of their attacks and minimize their likelihood of being detected. Our proposed trust model is feasible with the introduction of artificial intelligence and the central controller (C <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">c</inf> ) in 5G to support the hybrid trust model. We have explained in detail our proposed model that reinforcement learning based hybridization of trust model can enhance the performance of the network interms of learning and tackling intelligent attacks, whereby achieving context awareness and detection of malicious entities.