Research On Key Technologies Of Fault Tolerance In Wireless Sensor Networks

Fault tolerance is one of the critical issues in Wireless Sensor Networks(WSN)applications.The problems of node failure,communication link failure and data loss is inevitable in WSN.The fundamental reason is that there are many uncertainties in the deployment,communication,power supply,data collection and perception in WSN.Besides,WSN is vulnerable to interference of node energy,computing performance,working environment and other factors.If the node failure and network link loss occur in the actual application scenarios,the work efficiency,security and service quality of the entire network will be affected.Therefore,it is worthy concern how to eliminate the influence of fault and improve the fault-tolerance of WSN.The research of fault-tolerant key technology in WSN has important theoretical significance and great practical value.The reason of wireless sensor network failure may be the failure of deployed nodes,link error,network congestion,and the base station error,which involves network deployment,routing and data processing.In order to improve the ability of fault-tolerant and robustness,this paper conducts a fully research on the fault-tolerant technology of WSN in networok deployment,routing and data processing on the basic of spatial physic redundancy such as sensor nodes and information redundancy such as redundant data received from nodes.The research contents include the following four aspects:(1)research on fault-tolerant mechanism of WSN based on topology control;(2)research on fault-tolerant mechanism of WSN based on routing redundancy;(3)research on faulttolerant mechanism of WSN based on exception data repair;(4)research on fault-tolerant mechanism of WSN based on redundant data clustering.The main work and contributions are as follows:First of all,to address computational complexity and serious energy consumption of WSN faulttolerant topology control technology based on k-vertex connectivity,this paper proposes a topology fault-tolerant control a novel method for distributed WSN based on optimal control of independent path lifetime: DPLTO.DPLTO has two phases: initialization phase and connection restore phase.In the former phase,it constructs an initial optimized WSN topology and prepares for minimizing the amount of network resources when a high-quality recovery path is reached.In the latter phase,DPLTO introduces path tolerant failure time mechanism.It can reduce the routing information interaction by dynamic adjustment of path tolerance time.Thereby it further reduces the energy consumption during the chain reaction caused by the fault node.This mechanism can also effectively solve the problem of low fault tolerance when multi-node fault simultaneously.In addition,to optimize the energy consumption and topology of the network,DPLTO dynamically adjusts the transmission power of sensor nodes,which prolongs the life cycle of WSN.At the same time,DPLTO considers energy consumption from the local point of view.That is,DPLTO takes into account the balance energy consumption,computational complexity and network recovery efficiency of the WSN topology path.According to simulation results,DPLTO can effectively guarantee the k-vertex connectivity of the network,thus prolonging the life cycle of the whole network.Secondly,to solve the problem of the excessive energy consumption caused by redundant nodes in WSN routing redundancy design,this paper proposes a WSN fault-tolerant method BRFTR based on routing redundancy,which achieves the purpose of network energy saving.BRFTR introduces two important metrics: node energy consumption degree and node survival degree.The energy consumption degree metric can be calculated by using node working time and energy consumption,while node survival degree can be calculated by neighbor node life and energy consumption degree.In BRFTR,the ant colony heuristic algorithm is used to control the alternation routing redundant state between sleeping and working,which leads to the energy consumption of the routing redundancy is controlled in a reasonable range and prevent the performance degradation of WSN network due to premature failure of some key nodes.According to experimental simulation and results analysis,it can be seen that BRFTR can not only effectively control the energy consumption of gateway nodes,but also prolong the life cycle of gateway nodes.Therefore,BRFTR improves the availability of the network.Afer that,to deal with abnormality such as data loss,data error and invalidate data in the process of data collection when WSN nodes or WSN link occur an error,this paper proposes a new method ELOFGM which contains two stages: outlier detection phase and outlier repair phase.In the first phase,ELOFGM can effectively identify many kinds of abnormality such as loss data,error data and invalidate data by calculating the ratio of standard deviation of all data points in two adjacent time windows without the globle information.In the second phase,the GM(1,1)model for redundant data is constructed by using gray level theory.According to experimental simulation and results analysis,it can be seen that GM(1,1)can effectively complete the repair of abnormal data and improve the quality of data.In the last,the deployment of high-density distributed sensor nodes may lead to high correlation and redundancy,which leads to high redundancy in WSN aware data.To handle the issure,this paper proposes a new method MDRR-PCA based on R-PRC by using redundant data and k-means clustering.Since the extraction of the principal component transformation base of the R-PCA method is implemented in a recursive way,the MDRR-PCA can adapt to the dynamic change of WSN.MDRR-PCA has two characteristics:(1)it can effectively tolerate fault outliers by directly operating the cluster heads in WSN;(2)it can successfully apply the theory that the correlation of sensor data with the same physical variables of neighbor nodes is higher than the data of different physical variables collected from the same node.According to experimental simulation and results analysis,it can be seen that MDRR-PCA has better performance in fault detection and data recovery.Compared with the state-of-the-art methods,MDRR-PCA improves the accuracy of fault detection by about 20% and reduces the error of data recovery by about 28%.In summary,this paper systematically studies the fault-tolerant of WSN in network deployment,routing and data processing.Based on redundant nodes and redundant data,fault-tolerant mechanism for connectivity and transmission data validity of WSN has been studied.Fault-tolerant mechanisms are proposed for network topology,network routing and data transmission in the network respectively.According to the simulation experiment and result analysis,the validity of the proposed methods in the fault detection and fault tolerance are proved.These methods enhance the robustness,prolong life cycle and lay a solid foundation of WSN to provide better network services.