Optimal Selection of Power Saving Classes in IEEE 802.16e

The new IEEE 802.16e standard introduces two types of sleep modes for energy-efficient operations: power saving classes (PSCs) of type I based on binary-increasing sleep window size and PSCs of type II using constant sleep window size. This paper determines the optimal sleep mode selection for IEEE 802.16e by using the semi-Markov decision processes (semi-MDP). By means of evaluating corresponding cost metrics on system energy consumption and delay performance, we highlight the energy-performance trade-offs among different operational modes and formulate it as probabilistic constrained policy optimization (PO) problems. Our main goal is to search the space of all policies and to find the optimal one that achieves the minimum energy cost or traffic delay under different traffic requirements. Numerical results demonstrate the effectiveness of our semi-MDP method. We also investigate how the energy cost, delay penalty, as well as user objective jointly affect the space of the optimal policies.