WSN中传感器节点的弹性神经网络任务分配方法

Task Allocation Method of Sensor Nodes Based on MEMSOM Neural Network in WSN

为解决WSN多目标跟踪节点任务分配的竞争冲突问题,提出一种融合了模糊聚类的多弹性子模自组织神经网络节点任务分配方法.通过模糊聚类估计目标数量,建立节点任务分配跟踪精度和能量消耗的综合性能指标,采用非全连接的环形弹性结构自组织神经网络优化监测联盟,用最近邻法对神经元弹性子模进行初始化,根据胜者为王原则动态调整子模的感受域,以快速锁定最优监测联盟,实现多目标的精确跟踪.实验结果表明：文中方法能有效解决多目标跟踪节点任务分配的竞争冲突问题,以及竞争冲突时的系统能耗增加与实时性问题;在随机均匀部署节点拓扑和目标直线运动模式下,文中方法的能耗较最近邻法降低了48.2%~55.9%,较未改进弹性神经网络法降低了37.4%~42.5%,且运算速度提高了19.0%~27.4%.

In order to reconcile the competition conflict in the task allocation of sensor nodes during the multi-target tracking in wireless sensor networks（WSN）,a novel task allocation method is proposed based on the fuzzy clustering and the multiple-elastic module（MEM） self-organizing neural network.In this method,first,the target number is estimated via fuzzy clustering.Next,an integrated performance index is presented to describe the tracking precision and the energy consumption in the multi-target tracking.Then,a self-organizing neural network with non-fully-connected elastic neuron-ring structure is constructed.Moreover,the elastic submodule of neurons are initia-lized by means of the nearest neighbor method,whose receptive field is finally adjusted dynamically according to the WTA（Winner Take All） principle so as to quickly lock the optimal node alliance and ensure an accurate multi-target tracking.Experimental results show that the proposed method effectively reconciles the competition conflict in the task allocation of sensor nodes during the multi-target tracking,avoids further increase in energy consumption and prevents any possible real-time performance degradation due to the competition conflict;and that,in the conditions of random uniform node topology and linear target movement,the proposed method is superior to both the nearest neighbor method and the conventional MEM neural network method,respectively with an energy consumption decrease by 48.2%~55.9% and 37.4%~42.5%,and with a calculation speed increment of 19.0%~27.4%.