A Resource Allocation Scheme of D2D Energy Harvesting Networks Based on Stochastic Learning

Abstract

As one of the key technologies in the 5th generation mobile networks, device-to-device (D2D) can significantly improve spectrum efficiency (SE), throughput, energy efficiency, and reduce congestion in cellular networks. However, due to the spectrum sharing between D2D users (DUs) and cellular users (CUs), unreasonable resource allocation may cause serious interferences. Meanwhile, considering the energy limitation of D2D equipment and the requirement of green communication, this paper studies the resource allocation scheme under D2D energy harvesting (EH) networks based on stochastic learning and Stackelberg game. We formulate the joint channel and EH time slot allocation problem as a distributed learning problem. The DUs should pay for reusing the spectrum resource to base station if joining the networks reduce the SE of CUs severely. The proposed resource allocation scheme can converge to a Nash Equilibrium which maximizes the SE of the D2D networks and protects the quality of service of CUs. Finally, numerical results validate that the proposed scheme achieves an advanced performance relatively.