中文摘要：

针对机会网络中多副本报文转发机制下节点缓存溢出导致的拥塞现象,提出一种基于元胞学习自动机的拥塞控制策略。根据报文所在节点的局部环境中周围邻居节点对该报文的持有情况,按照给定的元胞规则对报文的丢弃概率进行自动学习及更新。在节点间进行报文复制时考虑对端节点上缓存报文的缓存熵信息,然后结合报文在当前节点的丢弃概率及邻居节点的缓存熵信息,对报文进行排序和丢弃。实验仿真结果表明,该策略有效降低了网络负载率和报文投递延时,并提高了报文投递成功率。

英文摘要：

In order to improve the throughput of opportunistic networks during congestion phase caused by multiple copies packet forwarding method,based on cellular learning automataa novel congestion control strategy was proposed. Different from conventional congestion control strategies,in which only particular information of nodes or packets are considered,this novel strategy takes into account the packets retain information from neighbor nodes. Each node is described as a cellular equipped with multiple learning automata in the network. According to the packets information stored in neighbor nodes,each node updates drop probability of packets under the rule of learning automata automatically. Furthermore,the buffer entropy of each neighbor node is taken into account when a packet is replicated,and a novel policy of dropping and replicating packets is also employed to increase nodes＇ entropy. The simulation results showed that the present approach effectively reduces the network overhead,packets delivery latency and improves packets delivery ratio.