State Space Recursive Least-Squares Based Adaptive Sampling for Sensor Networks

To improve the availability of communication bandwidth in distributed systems, communication overhead should be reduced as much as possible. This paper focuses on distributed data-stream systems. In such a network, large number of sensors delivers continuous data to a central server. The sampling rate of each sensor affects the communication resource and the computational load at central server. In this paper a new method for adaptive sampling is proposed, which is a modification of an existing adaptive sampling technique using Kalman filter. In this method the sampling rate at each sensor adapts to the streaming-data characteristics. This new approach employs State Space Recursive Least Square (SSRLS) based estimation technique wherein the sensor can use the estimation error to adaptively adjust its sampling rate within a given range autonomously. When the desired sampling rate violates the range, a new sampling rate is requested from the server. The server allocates new sampling rates under the constraint of available resources such that SSRLS estimation error over all the active streaming sensors is minimized. The effectiveness of the algorithm is demonstrated by presenting a comparison of SSRLS and uniform sampling approach, which is already proven flexible and effective.