Distributed Recursive Filtering Over Sensor Networks Under Random Access Protocol: When State Saturation Meets Censored Measurement
研究了传感器网络中状态饱和时变系统在测量删失下的分布式滤波问题，采用Tobit模型描述删失测量，并通过随机访问协议减少数据碰撞，设计了保证滤波误差协方差上界并最小化的滤波器。
In this article, a new distributed filtering problem is studied for a class of state-saturated time-varying systems over sensor networks under measurement censoring, where the censored measurements are described by the Tobit measurement model. To curb the data collision and ease communication burden, a random access protocol (RAP) is implemented onto the sensor-to-filter channels to orchestrate the transmission sequence of multiple sensor nodes. The purpose of the addressed problem is to construct a state-saturated distributed filter such that upper bounds (on filtering error covariances) are guaranteed and filter parameters are determined to accommodate both measurement censoring and state saturation under the RAP. By means of matrix difference equations, the desired upper bounds are first acquired and later minimized through appropriately designing filter parameters. Particularly, the sparsity issue with respect to the network topology is tackled via the employing certain matrix simplification technique. A simulation example is finally presented to showcase the applicability of the proposed state-saturated distributed filtering algorithm.