移动自组网自私行为闭环惩罚模型设计

Design of closed-loop model on selfish behavior in MANET

针对移动自组网MAC层开环模型缺乏自私行为惩罚机制的不足,设计一种包括报文信息统计、统计信息共享和判决惩罚模块的闭环改进协议PDCF。报文信息统计定义转发率、丢弃度等指标衡量节点行为,修改仿真软件内置函数实现节点行为信息的统计。统计信息共享定义行为量化等级,改进DCF预留字段和CSMA/CA机制,实现节点行为信息的全网覆盖。判决惩罚模块对节点行为建立一维马尔可夫链惩罚模型,限制节点自私行为。仿真结果表明,与DCF相比,PDCF时延、丢包率和控制开销基本不变,但稳态时自私节点比例得到有效控制,且吞吐量增加2.7 Mbps,网络性能得到改善。

This paper proposed a new closed-loop protocol named penalty distributed coordination function （PDCF） to im- prove the deficiency about recent open-loop model, which lacked penalty mechanism in medium access control （ MAC ） layer of mobile Ad hoc networks （MANET）. PDCF included three modules, packets statistics （PS）, information sharing （IS） and behavior verdict penalty （BVP）. First in PS, it defined forwarding rate and dropping level to judge node behavior, and col- lected information by modifying inline function of simulation software. Then it delivered collected information about nodes to oth- er nodes by IS, which defined quantization level and improved reserved fields as well as CSMA/CA mechanism. At last deve- loped a one-dimensional Markov chain penalty model to limit the node behavior based on previous information. Results show that, delay, packet drop rate and control overhead of PDCF were similar with that of DCF. However, proportion of selfish nodes is lower and throughput increases about 2.7 Mbps in PDCF, all of which are better than those of DCF.