EFFICIENT RESOURCE ALLOCATION IN ENERGY HARVESTING WIRELESS NETWORKS
By: NEYRE TEKBIYIK ERSOY
Recent advances in the area of energy scavenging enable communication devices that may harvest energy through solar cells, vibration absorption devices, thermoelectric generators, wind power, etc. However, having a renewable energy supply feature calls for novel design principles to efficiently utilize the dynamic levels of available energy. Hence, energy harvesting shifts the nature of energy-aware solutions from minimizing energy expenditure to optimizing it over time. There has been considerable recent research effort on optimizing data transmission with an energy harvesting transmitter. The work in this thesis differs from such previous studies particularly in its aim to maximize the throughput in a proportionally fair way, taking into account the inherent differences of channel quality among users. We investigate the proportional fairness based utility maximization problem in a time-sharing multiuser Additive White Gaussian Noise (AWGN) broadcast channel, where the battery of the transmitter gets recharged periodically. As time is shared among users, the optimal solution of the short-term proportional throughput maximization problem presented in this thesis entails not only the optimal power/rate allocation to users between energy arrivals, but also the optimal time allocation that will maximize the proportionally fair utility function. The structural properties of the problem were analyzed, and its biconvexity has been proven. Due to the biconvex nature of the problem, a Block Coordinate Descent (BCD) based algorithm has been developed to obtain the optimal solution. The characteristics of the optimal solution have also been derived. Thus, two simple and computationally scalable heuristics, PTF and ProNTO, are proposed.
Energy harvesting, optimization, block coordinate descent, biconvex, proportional fairness.