Power-Fractionizing Mechanism: Achieving Joint User Scheduling and Power Allocation via Geometric Programming

In the 4G and forthcoming 5G, the throughput is largely improved. Let edge users be the users beyond a certain distance to the associated BS in their covering cell, where the receiving intended signal is relative small. Throughput of edge users is the key metric measuring system performance. Originally designed for improving the performance of edge users under different scenarios, many technologies of user scheduling and power allocation, such as coordinated multipoint and Nash noncooperative power game are proposed. However, the existing algorithms are generally heuristic leading to a large performance gap in different scenarios. In this paper, we formulate a joint optimization to integrate user scheduling and power allocation into a unified optimization framework. By proposing a novel power-fractionizing mechanism, binary optimization can be avoided and the scheduling can be incorporated into the power allocation process. Thereafter, the joint optimization problem is then transformed into a signomial programming (SP) problem, which is further approximated by iteratively solving a series of tractable geometric programming problems that can converge to at least a local optimum of the SP problem. Unlike the existing heuristic algorithms dedicated to specific scenarios, our algorithm is more robust and environment adaptive. Simulation results show that the proposed joint optimization outperforms the existing heuristic approaches by significantly improving the total edge user throughput with stable performance.