METHOD FOR MEASUREMENTS IN WIRELESS SENSOR NETWORKS WITH SENSORS WITH SLOW DRIFT OF PARAMETERS

Abstract

A method for increasing the accuracy of measurements in wireless sensor networks has been developed. The method is based on the detection of errors and correction of measurement results. The corrected results are used to filter and predict the drift of the sensor parameters. Structures and principles of construction of mobile agents in hardware and software representation are analysed. Advantages of group application of agents are: a greater range of action, achieved through the dispersal of agents throughout the work area; an extended set of executable functions, which is achieved by installing on each agent of individual executing devices; higher probability of a task accomplished by the possibility of redistribution of goals among group agents in the event. The necessity of developing methods for automatic calibration of sensors in wireless sensor networks is substantiated. To solve this problem, a method of iterative evaluation and correction of measurement results is proposed. There are offered the mathematical models of slow sensors parameters drift in specialized wireless sensor network, which have a lot of sensors and one sink point. At the mutual correction of sensors parameters drift the errors of measuring are reduced and the life cycle of the effective functioning of network is prolonged. It is especially important for networks, access to which is difficult for some reasons. The method of errors detection and correction of results of sensors measuring in wireless sensory networks is offered. The method is based on supposition about spatio- temporal correlation of measuring results in nearby sensors and absence of cross-correlation of technical parameters drift.