基于部分可测马尔科夫决策过程业务感知的微基站休眠时长确定策略

Micro Base Station Sleeping Cycle Determination Strategy Based on Partially Observed Markov Decision Process Traffic Aware

针对密集组网场景中业务不确定性引起的基站休眠周期难以确定的问题,该文提出一种基于部分可测马尔可夫决策过程(Partially Observed Markov Decision Process,POMDP)业务感知的微基站休眠时长确定策略。该策略将周期分为长周期和短周期,每个周期由轻度和深度两个阶段构成。通过POMDP感知到达基站的业务状态,动态调整周期时长,进而选取适合当前周期的时长。仿真结果表明,该策略可以根据业务感知提前确定微基站关断时长,与基于业务门限值的基站关断机制相比节能效果更好。

In order to solve the problem that the sleeping cycles are difficult to be determined duo to the traffic uncertainty in dense network scenarios, this paper proposes a Micro base station sleeping cycle determination strategy which based on the Partially Observed Markov Decision Process （POMDP） traffic-aware. In this strategy, the sleeping cycle is divided into long cycle and short cycle, and each cycle consists of deep and light stage. Based on the POMDP traffic-aware, it can dynamic adjusting the cycle and determine the proper length of cycle. Both the analytical and simulation results show that compare with sleeping strategy based on the traffic threshold, the base station sleeping strategy based on traffic awareness can effectively reduce the energy consumption of the micro base stations in the dense network by adjusting the sleeping time of the micro base stations in real time.