Joint Computation Offloading, Mode Selection and Resource Allocation for D2D-Assisted Mobile Edge Computing

The emergence of 6G technologies has accelerated the development of advanced intelligent applications, leading to an increased demand for mobile edge computing (MEC) systems. However, limited computational resources result in a reduced quality of service (QoS) and inefficient resource utilization. To address these challenges, we propose a device-to-device-assisted MEC (D2D-MEC) architecture that optimizes the joint allocation of communication and computational resources. The objective is to minimize the average task offloading cost, which is defined as a weighted sum of delay and energy consumption. This goal involves the joint optimization of computation offloading, communication mode selection, and resource allocation. By framing the problem as a Markov decision process (MDP), the algorithm implements a centralized training and decentralized execution approach. This approach enables agents to dynamically adjust their offloading strategies, communication modes, and resource allocations based on real-time network conditions and task requests. Simulation results indicate that the proposed method achieves higher system rewards and lower average offloading costs compared to baseline schemes.