6LoWPAN中分布式TCP缓存队列策略  

DISTRIBUTED TCP CACHING QUEUE STRATEGY IN 6LOWPAN

作　　者：吉福生[1] 周小龙[1] 

机构地区：[1]重庆邮电大学宽带泛在接入技术研究所,重庆400065

出　　处：《计算机应用与软件》2016年第5期120-123,131,共5页Computer Applications and Software

基　　金：国家自然科学基金项目(61371097)

摘　　要：在多TCP连接的6Lo WPAN(IPv6 over Low-Power Wireless Personal Area Networks)网络中,针对丢包严重时存在频繁的端到端重传问题,提出分布式TCP缓存队列策略。该策略能够使中间节点合理地缓存不同TCP连接在链路层传输中丢失的分段,从而保证每个TCP连接的性能,减少网络能耗。另外,采用ARQ机制进行链路层数据帧传输时,由于中间节点判断丢包的准确率较低,导致缓存队列中存在一些实际未丢失的分段。对这些分段的重传会消耗额外的能量,降低了缓存队列的利用率。因此缓存管理采用询问邻居节点的方式检查分段是否丢失,及时地删除无效的缓存分段。实验结果表明,采用分布式TCP缓存队列策略可以使得多个TCP连接的网络性能以及缓存队列利用率得到了很大提高。We proposed a distributed TCP cache queue strategy for the problem of frequent end-to-end data retransmission when the packets＇ missing is serious in 6LoWPAN with multiple TCP connections. This method can make the intermediate nodes reasonably cache the missing segments of different TCP connections when the link layer transmission is failure,so as to ensure the performance of each TCP connection and reduce the energy consumption of the network. In addition,when transmitting the data frames of link layer by adopting ARQ mechanism,there are some actually non-missing segments in cache queue due to the intermediate nodes having low accuracy in judging the packet missing,to retransmit these non-missing segments will cost extra energy and reduce the utilisation of cache queue. Therefore,the cache management checks whether or not the segment has lost by asking the neighbouring node and deletes the invalid cached segments timely. Experimental results showed that the performance of the network with multiple TCP connections and the utilisation of cache queue had been greatly improved by using distributed TCP cache queue strategy.

关 键 词：低功耗个人局域网 传输控制协议 自动重传请求 缓存队列 丢包恢复 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]