Event-Triggered Distributed Estimation with Decaying Communication Rate

We study distributed estimation of a high-dimensional static parameter vector through group sensors whose communication network is modeled by fixed directed graph. Different from existing time-triggered schemes, an event-triggered asynchronous scheme investigated in order to reduce while preserving convergence. A algorithm with single step size first proposed based on time-dependent decaying threshold. With the scheme, each sensor sends its estimate neighbor only when difference between current and last sent-out larger than triggering prove that has mean-square almost-sure convergence respectively, under integrated condition topology measurement matrices. The satisfied if balanced digraph containing spanning tree system collectively observable. Moreover, we provide estimates for rates, which are related as well Furthermore, essential metric intensity algorithms, rate proved decay zero certain speed almost surely time goes infinity. show given size, adjusting threshold can lead tradeoff error rate. Specifically, increasing would make faster, but error. Numerical simulations provided illustrate developed results.