Dynamic Satisfactory Power Allocation for Multi-Class Ultra Dense 5G and Beyond

Power control is originally used in wireless networks, to compute optimal transmit power and deal with undesired interference. It is also a flexible mechanism that can provide Quality of Service (QoS) and strategically allow to meet the user requirements. This paper introduces a QoS-aware Satisfactory Power Allocation (SPA) for ultra dense networks, using a mean field perspective. In this setting, the user devices are partitioned into several classes based on their throughput requirements. Now, instead of seeking to maximize their QoS, the user devices from each class only aims to meet their respective throughput demands. Yet, by leveraging stochastic geometry analysis and mean field approximation, we investigate the uplink power control problem in a large scale ultra dense network that guarantees a satisfactory performance per class. Next, we formulate the problem as a mean field optimal control where the optimality conditions are derived using Lagrangian dual formulation. Finally, the effectiveness of the SPA policies is illustrated via extensive numerical analysis, and many insightful discussions are presented.