Research And Implementation Of Clustering Algorithm For Distributed Spectrum Monitoring Network

With the development of wireless communication technologies,the electromagnetic spectrum resources become increasingly scarce.Therefore,it is important to monitor the distribution and use of the spectrum resources in both civil and military fields.Due to the limitations of the monitoring scope,monitoring bandwidth and layout environment,the traditional single node monitoring can only process a narrow band of electromagnetic signals within the local area.In recent years,wireless sensor networks(WSNs),with the characteristic of self-organization,strong robustness and so on,are widely used in spectrum monitoring field.Therefore,the network monitoring is becoming a new development trend of the electromagnetic spectrum monitoring.To reduce the network energy consumption and prolong the network lifetime,grouping sensor nodes into cluster has been widely used in wireless sensor networks.As the electromagnetic spectrum monitoring network is a special wireless sensor networks,clustering is not only to prolong the network lifetime but also to serve for high precision positioning.Therefore,this thesis mainly studies the clustering algorithm for location in different application scenarios.The main research connects are as follows:1.Aiming at the problem of clustering for prolonging the network lifetime and serving for high precision positioning in fully connected wireless sensor networks with a single target and a huge number of sensor nodes deployed randomly,a new clustering algorithm for single target locating is proposed.In order to balance the network energy consumption and location accuracy,the discrete particle swarm optimization is used for selecting the optimal node sets.Furthermore,a new clustering process is designed to group the optimal node sets into cluster for further location.The proposed algorithm is compared with HEED under the RSSI/TDOA two rounds cooperative location scheme.Simulation results show that this algorithm improves the location accuracy with low energy consumption.2.In order to prolong the network lifetime and serve for high precision positioning in multi-hop wireless sensor networks with a single target and a huge number of sensor nodes deployed randomly,an optimal node sets selection algorithm is proposed based on the improved discrete particle swarm optimization algorithm by trading off the network energy consumption,the energy load balance,location accuracy and network connectivity.Furthermore,a new clustering algorithm is presented to group the optimal node sets into cluster for further location.The simulation analysis is carried out under the RSSI/TDOA two rounds cooperative location scheme.Simulation results show that this algorithm guarantees the member nodes multi-hop communication in a cluster,prolongs the network lifetime and improves the location accuracy.3.Research from the aspect of realizing the clustering algorithm for location and avoiding the problem of high energy consumption of nodes covered by multi targets in the large scale wireless sensor networks with multi targets and sensor nodes deployed randomly,a new clustering algorithm based on evolutionary game theory(EGT)is proposed.The non-cooperative game theory model is established by making the mapping of the search space of the optimal node sets to the strategy profile space of game;then the optimization objective is achieved by using Nash equilibrium and the perturb-recover process of equilibrium states.Furthermore,a detailed clustering algorithm is presented to group the optimal node sets into cluster for further location.The proposed algorithm is compared with the nearest neighbor algorithm and the clustering algorithm based on discrete particle swarm optimization algorithm in the location accuracy and the network lifetime under the RSSI /TDOA two rounds cooperative location scheme.Simulation results show that the algorithm avoids the problem of high energy consumption of the nodes covered by multi targets,prolongs the network lifetime and guarantees the location accuracy.4.In order to realize the simulated training project for the census and supervision of the spectrum resources,the location and tracking of the main target and so on,a distributed electromagnetic spectrum monitoring simulation training system is constructed based on HLA(High Level Architecture)using Visual Studio2003,Oracle9 i,MapInfo and Matlab7.0 as the programming tool.The clustering algorithms proposed above are implemented in the simulation training system.Furthermore,the efficiency of the clustering algorithms is proved.Since the training system has strong expansibility,it can provide a platform for researching other key technologies in distributed electromagnetic spectrum monitoring network.