基于低秩矩阵恢复的移动WSN节点轨迹拟合研究

对传感节点的位置和轨迹信息进行更新和管理,是传感节点可移动的无线传感器网络系统的主要特征。传感节点的位置和轨迹信息频繁传输会增加网络的能量消耗。为了降低信息的传输量,对信息进行采样,并通过拟合传感器节点的移动轨迹恢复原始轨迹信息;为了进一步提高拟合准确度,将压缩感知理论应用于轨迹拟合中,该算法对非凸最优化问题进行松弛,将矩阵的秩松弛到矩阵的Frobenius范数,并转化为非约束优化问题,然后采用最小二乘法对目标函数进行迭代以求得最优解。仿真实验结果表明,算法能够较好地拟合传感节点的移动轨迹,能显著减少传感节点位置和轨迹信息的发送量。

Research on Path Fitting of Mobile Nodes in Mobile WSN Based on Low-rank Matrix Recovery

Updating and managing the position and path information of mobile sensor nodes, which is one of the main features of mobile wireless sensor network (MWSN) system. Frequently transferring the path information will increase the energy consumption of MWSN. In order to reduce the transmission of information, we apply the algorithm based on low-rank matrix recovery on path fitting. The algorithm relaxes the rank of matrix by replacing it with the Frobenius norm to simplify the non-convex optimization problem, turns the problem into the non-constrained problem, and uses an alternating least squares procedure to find the solution. Experiment results show that the algorithm achieves good accuracy on path fitting and reduces the transmission of path information in the meanwhile.