适用于无线传感器节点的低复杂度CCVSLMS均衡器研究

摘要

针对无线传感器节点的均衡算法--最小均方误差(LMS)算法和截断数据LMS算法或截断误差LMS算法(CLMS算法)的收敛速度、稳态误差性能的不足,进行了截断误差和截断数据可变的步长LMS算法(CCVSLMS算法)的研究,该算法通过对误差和数据进行截断处理和建立步长因子μ与截断变化率的线性函数关系来改善收敛速度和稳态误差之间的矛盾,并有效降低了计算复杂度.仿真结果证实该算法性能优于LMS算法和CLMS算法.在TSMC 90nm CMOS工艺下,均衡器的超大规模集成(VLSI)优化的综合结果显示,硬件实现面积约为1806个单元,动态功耗仅为12μW,因此该均衡器能满足无线传感器节点高灵敏度和低成本的要求.%In view of the poor performance of steady state error and convergence speed of the least mean squares (LMS) algorithm and the clipped-data LMS or clipped-error I1S (CLMS) algorithm for equalization of wireless sensor nodes, this paper proposes a clipped-data and clipped-error variable step size LMS (CCVSLMS) algorithm. The algorithm carries out clipped-data and clipped-error and builds a linear function relationship between the step size factorμ and the change rate of sign, so it could decrease the contradiction between the convergence speed and the steady state error, and had a less computational complexity. The simulation results confirm that the performance of the algorithm surpasses the LMS algorithm and the CLMS algorithm. Moreover, the digital circuit of the equalizer was designed optimally. In the TSMC 90nm CMOS process, the VLSI synthesis results have shown that the dynamic power is only 12μW andthe cells area is about 1086 by the design compiler (DC) tool. Therefore, the equalizer can satisfy the high sensitivity and lower cost of wireless sensor nodes.