Reliable Kalman Filtering with Conditionally Local Calculations in Wireless Sensor Networks

Wireless sensor networks state assessment is one of the areas of research in digital signal processing. Traditional algorithms include centralized and distributed filtering of data received from sensors. These algorithms iteratively use the information obtained in the course of measurements from all pairs of sensors, leading to an increase in the computational load and a decrease in algorithm reliability. This article proposes an algorithm for distributed reliable filtering with conditionally local aggregation of data received from sensors for a wireless sensor network to solve this problem. Software simulation has shown the possibility of minimizing the upper bound of the mean squared error update error that occurs when processing a noise faulty communication channel compared to known algorithms. The ability to use information from neighboring pairs of sensors and local measurements in the proposed algorithm made it possible to accelerate the appearance of stability of the value in errors. It is proved that the algorithm proposed in the paper is scalable for large networks. The results can be effectively applied in various wireless monitoring systems.