Energy efficient resource allocation and admission control for D2D-aided collaborative mobile clouds

In this paper, we investigate the joint admission control and resource allocation problem for collaborative computation under fading channels. We develop an Internet-of-Things (IoT) framework in which establishing Device-to-Device (D2D) communications, resource-poor wearable Source Mobile Terminals (SMTs) may offload their computations to resource-rich Processing Mobile Terminals (PMTs), or execute them locally, so as to save energy. Considering the offloading scenario, first, a probabilistic admission control algorithm is proposed for Mobile Terminals (MTs) taking both the deadline and energy harvesting constraints into account. Then, the joint CPU clock frequency/transmit power allocation and collaborative pair selection problem for MTs is addressed mathematically. For local execution scenario, optimal CPU clock frequencies are obtained for SMTs. Finally, based on energy consumption and outage imposed by each scenario, SMTs decide whether to offload their computations or execute them locally. Simulation results demonstrate that the proposed D2D-aided Collaborative Mobile Cloud (DCMC) approach attains a near-optimal energy expenditure in a semi-feasible system while effectively mitigating outage ratio of MTs.