基于牛顿迭代二值采样信号参数估计的鲁棒算法

Robust Algorithm for Parameter Estimation of Binary Sampled Signal Based on Newton’s Method

无线传感器网络中,低比特数据传输和处理有利于节约网络的开支和节点能耗。然而,低比特数据在传输过程中,很容易受到干扰而导致比特值随机反相。论文研究利用非理想信道情况下接收到的1比特采样信号的变量含误差(EIV)模型参数估计问题,提出了一种基于牛顿迭代的鲁棒估计算法。该算法首先将问题转化为基于比特扰动概率的极大似然优化问题,然后利用牛顿迭代算法获得问题的最优解。该算法有较快的收敛性,同时针对EIV模型中的乘性噪声和比特值随机反相问题有较好的鲁棒性。论文推导了克拉美罗下界(CRLB),并且进行了性能仿真,结果验证了算法的有效性。

In wireless sensor networks,low bit data transmission and processing can save network expenses and node energy consumption.However,low bit data is easy to be interfered in the process of transmission,which leads to the random inversion of bit value.We study the parameter estimation problem of the EIV(Errors-In-Variables)model of the received one-bit sampled signal in non-ideal channel,and propose a robust estimation algorithm based on Newton’s method.The algorithm first transforms the problem into a maximum likelihood optimization problem based on bit perturbation probability,and then uses Newton’s method to obtain the optimal solution of the problem.The algorithm has fast convergence.At the same time,the algorithm is robust to multiplicative noise and random inversion of bit values in the EIV model.In this paper,the Cramér-Rao lower bound is derived,and the simulation results verify the effectiveness of the algorithm.