Study Of A Data-sensitive Sensor Node Fault Detection Technology Under Severe Environment

Wireless sensor network is always deployed in danger area or the place that thepeople can’t get in to monitor the environment for the reason that Wireless sensornetwork can be used as event detection or target monitoring.Suffered from theenvironment influence, such as nosie and trend, sensor sampled data is inaccurate,and the feature is not obvious.At the same time the sensor cost much more energyand a large area of nodes may malfunction or perform badly. In harsh environment,if the wireless sensor network detect fault periodically, it can’t find the fault in timewhen sensor network goes wrong, what’s more, it will cost much more energy whenthere is no fault occurs. So wireless sensor network should be able to detect faulttimely, and those sensor nodes that perform badly shouldn’t be adopted like otherfault sensors which can’t be repaired, we should repair and go on making use ofthem by means of fault detection technology, so that we can reduce the cost andmanual labor of redeploying the sensor network. As a result, we should takedifferent technology for different fault type.According to the problem below, a data-sensitive sensor node fault detectionalgorithm is presented. Firstly this algorithm reduce the influence of badenvironment by the Resistant to environmental factors function of its ownanti-jammer and extract the data features to Enhanced time series features so that itcan reduce the communication traffic and energy cost. Secondly sensor is sensitiveto data and use hypothesis test to count the data collected in a period of time, sensorrecognizes the abnormal status of feature so that it can decide when to start the faultdetection. Finally sensor start fault detection, and take different method to detectdifferent type fault. It combines decision spread with vote decision to detect randomreading fault and adopt the fault sensor. To detect the drift reading fault, it makesuse of event detector and fault detection technology, and through retraining theevent detector sensor fixes itself to adapt the drift data and reduce the cost andmanual labor.