论文部分内容阅读
提出一种基于动态指数平滑模型对网络流量负载进行预测的负载均衡协议DES-LBPTP(a Dynamic Exponential Smoothing Model-Based Load Balancing Protocol for Traffic Prediction in Ad Hoc Networks).该协议以MAC层接口队列中分组长度为流量负载的衡量依据,利用动态指数平滑预测模型对节点的流量负载进行预测,根据预测到的下一时刻流量负载状况,在节点出现拥塞丢包之前,提前实施路由更新机制,避免中间节点拥塞,以此提高网络性能.此外,该协议在中间节点根据流量负载状况有选择地转发RREQ、在目的节点采用延时应答也在一定程度上改善了网络性能.仿真结果与AODV协议相比,端到端时延降低约50%,归一化开销改善28%,分组投递率最大增长10.4%.
A Dynamic Exponential Smoothing Model-Based Load Balancing Protocol for Traffic Prediction in Ad Hoc Networks (DES-LBPTP) is proposed based on dynamic exponential smoothing model. The protocol uses packets in the MAC layer interface queue Length is the basis of the traffic load, and predicts the traffic load of the node by using the dynamic exponential smoothing prediction model. According to the predicted traffic load status at a next moment, before the node experiences congestion and packet loss, the route update mechanism is implemented in advance to avoid the intermediate node Congestion in order to improve network performance.In addition, the protocol selectively forwards RREQ according to the traffic load status at the intermediate node, and delays the response at the destination node to a certain extent, improves the network performance.Compared with the AODV protocol, The end-to-end delay is reduced by about 50%, the normalization overhead is improved by 28% and the maximum packet delivery rate is increased by 10.4%.