Optimal Joint Base Station And User Equipment Admission Control For Green Wireless Cellular Networks

In the context of dramatically rising energy costs and carbon footprint of mobile networks, energy efficiency has become a key criterion in the design of future green wireless cellular networks. In a typical cellular network, base stations (BSs) are responsible for60-80%of the whole cellular network energy consumption. Moreover, a BS consumes more than90%of its peak energy even when there is little traffic. Therefore, to improve energy efficiency, it is crucial to reduce BSs’energy consumption in green wireless cellular networks.Dynamic base station (BS) operation (switch off or sleep mode) schemes and user equipment (UE) admission control are two important aspects in cellular networks in terms of energy efficiency and the quality of service (QoS). Although optimal policies for these two control schemes have been derived in previous works, these two important aspects have not been considered jointly. Considering these two processes jointly will be helpful to derive optimal solutions for both. Thus, a joint admission control for both BS and UE is proposed in this paper to obtain the maximum energy efficiency while guaranteeing the QoS. For energy efficiency, bit-per-joule(bits/J) is used as the performance measure. For user QoS, blocking probability is used as the performance measure. The tradeoff between energy efficiency and user QoS will be shown. Also, coordinated multipoint (CoMP) is used among the active BSs to improve user QoS in the cells whose BSs have been shut down to save energy. Furthermore, the problem is formulated as a semi-Markov decision process. An optimal solution can be obtained from a linear programming algorithm. The paper has shown that, different from the deterministic policies in existing works, the optimal BS operation policy is a randomized policy, i.e., a BS should be switched on/off with some probabilities when the system is in certain states.However, for a large network with many BSs and UEs, the linear programming approach can become computationally intractable. Therefore, structural results for this problem have been further presented in this paper by analyzing the system model. Future work is in progress to consider packet traffic characteristics and service latency in the proposed framework.