Analysis of the optimal sampling rate for state estimation in sensor networks with delays

Miguel Martínez-Rey; Felipe Espinosa; Alfredo Gardel

doi:10.1016/j.isatra.2017.03.007

Analysis of the optimal sampling rate for state estimation in sensor networks with delays

ISA Trans. 2017 May:68:293-301. doi: 10.1016/j.isatra.2017.03.007. Epub 2017 Mar 28.

Authors

Miguel Martínez-Rey¹, Felipe Espinosa², Alfredo Gardel³

Affiliations

¹ Electronics Department, Engineering School, University of Alcala, University Campus s/n, 28871 Alcalá de Henares, Madrid, Spain. Electronic address: miguel.martinez@depeca.uah.es.
² Electronics Department, Engineering School, University of Alcala, University Campus s/n, 28871 Alcalá de Henares, Madrid, Spain. Electronic address: felipe.espinosa@uah.es.
³ Electronics Department, Engineering School, University of Alcala, University Campus s/n, 28871 Alcalá de Henares, Madrid, Spain. Electronic address: alfredo.gardel@uah.es.

PMID: 28359530
DOI: 10.1016/j.isatra.2017.03.007

Abstract

When addressing the problem of state estimation in sensor networks, the effects of communications on estimator performance are often neglected. High accuracy requires a high sampling rate, but this leads to higher channel load and longer delays, which in turn worsens estimation performance. This paper studies the problem of determining the optimal sampling rate for state estimation in sensor networks from a theoretical perspective that takes into account traffic generation, a model of network behaviour and the effect of delays. Some theoretical results about Riccati and Lyapunov equations applied to sampled systems are derived, and a solution was obtained for the ideal case of perfect sensor information. This result is also interesting for non-ideal sensors, as in some cases it works as an upper bound of the optimisation solution.

Keywords: Kalman filter; Optimal sampling; Riccati equation; Sensor network; State estimation.