Ad hoc网络中Backpressure调度算法延迟性能的改进  

Delay improving of Backpressure scheduling algorithm on Ad hoc networks

作　　者：胡雪晴[1] 周继鹏[1] 陈晓霞[1] 

机构地区：[1]暨南大学信息科学技术学院,广州510632

出　　处：《计算机应用研究》2017年第6期1812-1816,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(61373125);广东省科技计划资助项目(2013B010401016)

摘　　要：为了成功传输数据包,Backpressure算法需要形成从源节点到目的节点的递减队列差梯度。但是,形成队列差梯度需要一段时间并且所需时间的波动性大,容易导致数据包平均延迟变大。通过分析时间波动性的原因,提出了改进的Backpressure算法(DH-BP),以解决在Ad hoc网络下有固定路由的Backpressure算法的延迟问题。DH-BP算法用由队列长度、队头延迟、路由跳数决定的计数器值代替队列长度值,来计算链路调度集合,从而实现链路调度。理论上证明了DH-BP算法具有throughput-optimal性能。仿真结果表明,DH-BP算法相比另外两种算法,其延迟性能和总吞吐量都有改善。In order to transmit data packets successfully, Backpressure algorithm needed to generate decreasing gradient of queue length difference from the source node to the destination node. However, the formation of gradient took a period of time which had large fluctuations, and easily led to packet average delay becoming large. By analyzing the reasons for time volatility, the paper proposed improved Backpressure algorithm associated with hops and delay（DH-BP） to solve delay of Backpressure algorithm for Ad hoc network scenarios with fixed routes. DH-BP algorithm used the counter value decided by queue backlog, head of line delay and route hops instead of the queue length value to calculate link scheduling＇ set and achieved link scheduling. The paper proves that DH-BP has the performance of throughput-optimal in theory. Simulation results show that DH-BP has better delay performance and total throughput compared to the other two algorithms.

关 键 词：Backpressure算法 AD HOC网络 最优吞吐量 延迟 链路调度 

分 类 号：TP393.07[自动化与计算机技术—计算机应用技术]