Performance Evaluation And Optimization For The Sleep Mode In WIMAX With Duplex Links

The energy consumption of battery-powered Mobile Station is one of the most significant issues in WiMAX. Sleep mode is introduced in IEEE 802.16e to save energy for Mobile Station. According to the cooperated operation mechanism of Mobile Station and Base Station, and the feature that the sleep mode can be interrupted by transmission of the uplink traffics, considering the influence of the duplex traffics on the system, and the traffic feature of different data frames in WiMAX, power saving class of type II and type III are modeled mathematically, analyzed theoretically and simulated systematically.Firstly, based on the work in principle of the power saving type II, which is recommended for Unsolicited Grant Service and Real-Time Variable Rate Service traffics, considering the feature that data frames can be transmitted in the listen state, the listen state for transmitting data frames is regarded as a listen busy period, the awake state is regarded as an awake busy period. Then, a queueing model with synchronous multiple vacations and two busy periods is built to capture the working principle of the power saving class of type II by taking into account the duplex links in WiMAX.Secondly, according to the work principle of the power saving typeⅢ, which is recommended for Multicast Connections and Management Operatons traffics, and the feature the no data frames can be trasmitted in the listen state, the whole listen state is regarded as a listen vacation period. Then, a queueing model with synchronous multiple vacations and one busy period is built to capture the working principle of the power saving class typeⅢby taking into account the duplex links in WiMAX.Thirdly, the queueing model with synchronous multiple vacations and two busy periods, and the queueing model with synchronous multiple vacations and one busy period are analyzed theoretically, respectively. By using the method of semi-Markov process and two-dimension embedded Markov chain, the performance measures are built, and expressions for the performance measure are given for these two models.Finally, the dependence relationships of the system performance measures and the length of the sleep window are discussed by numerical experiments. With the numerical results, the influence of the system load on the system performance is numerically evaluated. The theory analysis results are validated by simulation experiments. Considering the different influence of the the system measures on system performance, a cost function is constructed for the power saving class of typeⅡand typeⅢ, respectively. The optimal value of the length of the sleep window is given to optimize the performance of the system.