Edge Computing Resource Allocation Based on Trust and Multilateral Bargaining

Edge computing is important in advancing 6G networks by enabling low-latency, high-capacity data processing at the network edge. This paper presents an innovative approach for autonomously selecting edge servers and distributing computing resources to efficiently manage the users' offloaded tasks. A multi-server edge computing environment is introduced, capturing the users' computing and communication characteristics for effective task offloading. A reinforcement learning mechanism based on the Optimistic Q-Learning with Upper Bound Confidence Action Selection algorithm allows the users to autonomously select edge servers, considering their communication characteristics, the edge servers' computing capacities, and their trust levels. An innovative trust model quantifies the direct and indirect trust of the users in the edge servers considering their experience quality of service. Additionally, a multilateral bargaining mechanism allocates the edge servers' computing resources based on the users' service requests' order and their data processing needs. Simulation-based evaluations using real data demonstrate the proposed framework's operational characteristics and its superiority over existing mechanisms.