延迟容忍网络(DTN)是一种面向移动与极端网络环境的特殊无线自组织网络。相对于传统网络,DTN中节点需要长时间存储/携带消息,进而实现消息的转发,从而使得节点缓存成为影响网络路由性能的重要因素。为优化Epidemic路由协议中缓存管理机制,避免由活跃消息丢弃所引起的路由效率降低的情况,提出了一种基于消息生存属性的缓存管理(MPBBM)算法。该算法通过综合分析消息转发次数与生存时间等因素对消息传递的影响,制定了优化的缓存替换策略,使得缓存替换过程中有限保留新产生的消息、丢弃即将失效的消息。仿真结果表明,相比于其他缓存管理算法,MPBBM算法能够有效地提升消息交付率,并显著地降低投递时延与网络开销。
Unlike traditional networks, the nodes in delay tolerant network (DTN) may storage and carry messages for a long time and then forward them somewhere, which makes the buffer sizes of the nodes become an important factor of routing efficiency. For the purposes of optimizing buffer and replacement strategy and avoiding active messages discarded, a message-period-based buffer management (MPBBM) algorithm is proposed. According to the comprehensive analysis of forwarding amount and time to live (TTL) of the messages, MPBBM forms a buffer replacement strategy to make new messages stored and inactive messages dropped. The simulation results show that MPBBM, comparing with other algorithms, can improve the delivery rate and reduce the network overhead.