Cost Minimization for Energy-Constrained Partial Offloading in Cognitive Capacity Harvesting Networks

In this article, we investigate how to provide computation offloading service for secondary users (SUs) in cognitive capacity harvesting networks. In particular, standing on the secondary service provider (SSP)'s point of view, we consider the partial offloading and minimize the total cost of the SSP while satisfying the energy and delay constraints of SUs. We optimize all the related variables comprehensively, including the partial offloading factor, the transmit power, the cognitive radio (CR) router selection, the offloading spectrum block selection, and the computing resource allocation in both SUs and CR routers. We formulate the considered problem as a mixed integer nonlinear programming. We theoretically show that when the selections of the offloading CR router and offloading spectrum block for SUs are given, the optimal computing frequency in SUs and CR routers can be determined by the transmit powers and partial offloading ratios. Based on this, the considered problem is decomposed into a transmit power and partial offloading ratio optimization subproblem, an offloading spectrum block selection subproblem, and a CR router selection subproblem. For the three subproblems, differential evolution, more complex task first, and coalition game algorithms are proposed to solve them, respectively. Simulation results show that the proposed algorithms can greatly improve the system performance.