Performance bounds of distributed adaptive filters with cooperative correlated signals

In this paper, we studied the least mean-square-based distributed adaptive filters, aiming at collectively estimating a sequence of unknown signals (or time-varying parameters) from a set of noisy measurements obtained through distributed sensors. The main contribution of this paper to relevant literature is that under a general stochastic cooperative signal condition, stability and performance bounds are established for distributed filters with general connected networks without stationarity or independency assumptions imposed on the regression signals.创新点本文考虑一类基于最小均方算法(LMS)的分布式自适应滤波问题, 在传感器网络的连通条件和回归信号的联合激励条件下, 证明了算法的稳定性并给出了跟踪误差方差的上界。我们对回归向量不需要传统的独立性或平稳性等苛刻假设, 使得本文所建立的理论对反馈系统的应用成为可能。