An efficient algorithm for fault tolerance in multisensor networks

Due to environment noise or sensor-failure, the measurements might be susceptible to errors; and some of them may even appear contradictory. Combining the numerous sensor measurements to minimize the uncertainty and to improve the reliability and fault tolerance is a central theme of the multisensor network. In this paper sensors are separated into two types: usable sensor, useless sensor. Based on the statistics theory, this paper introduces a new efficient algorithm for the sensor integration. Given n sensors of which m sensors are usable sensors, in this paper an efficient O(n²) fault-tolerant algorithm is presented in this paper even when m is much less than n/2. The sensor is given a different weight grounded on the precision of sensor in the algorithm to improve veracity. The simulation indicates the algorithm is robust against the number of faulty sensors and noise.