A New Integration Method for Reliable Abstract Sensors

Data integration is a crucial issue in wireless sensor networks. This paper proposes a new fault-tolerant integration method-overlap algorithm. We additionally simplify Marzullo´s algorithm to adapt to sensor nodes. Those two methods are used to combine several interval estimates into a reliable interval output even when some of input are faulty. We simulate those two algorithms in Matlab respectively, compare their performance. The experimental results show that the width of output interval produced by overlap algorithm is less than that of Marzullo´s algorithm, thus it is more accurate.