一种基于卡尔曼滤波的分簇WSNs拥塞检测与控制方案

A Congestion Detection and Control Algorithm for Clustering WSNs Based on Kalman Filtering

针对分簇结构下无线传感器网络簇首节点因负载过大易产生拥塞问题,提出一种基于卡尔曼滤波拥塞预测与缓解算法CMKBO。该算法首先基于卡尔曼滤波理论,依据当前簇首缓存占用情况,预测下一时刻簇首队列长度,并结合簇首吞吐量对网络拥塞程度进行综合预判;当网络拥塞程度预测值超过设定阈值时,在簇内寻找一个最优节点协助簇首进行数据缓存和转发以达到控制和缓解网络拥塞;簇首启动拥塞控制时,协助节点依据当前簇首拥塞程度选择簇内转发或簇外转发,以实现不同情况下的拥塞控制。仿真实验结果表明:该算法能够较准确地预测未来簇首拥塞状况,能较好地缓解簇首压力,较CODA算法有更好的网络特性。

To solve the congestion problem of cluster head nodes in wireless sensor networks(WSNs)due to excessive loads,a congestion prediction and mitigation algorithm(CMKBO)based on Kalman filtering theory is proposed.The algorithm firstly predicts the queue length of a cluster-head at the next moment according to the current occupied buffer size at a cluster head based on Kalman filtering theory,and then analyzes and predicts the congestion level at a cluster head.When the predicted level of network congestion exceeds the pre-set threshold,an optimal node is selected in the cluster to assist the cluster head in data caching and forwarding to control and alleviate network congestion.As the congestion control method is running at the cluster head,the assisting node can select intra-cluster or out-cluster data forwarding according to the congestion level of the current cluster-head,which can achieve congestion control in different situations.The experimental results show that the proposed algorithm can accurately predict the congestion level for a cluster head and relieve the cluster head pressure.Moreover,compared with CODA algorithm,it has better network characteristics.