Several Studies On Three-dimensional WSN Nodes Location Algorithm Based On LSSVR

Wireless Sensor Networks has wide applications and shows great practical value in variousfields due to its low cost，distributed and self-organization. As the basis for many WSNapplications，localization is one of the most crucial problems. The thesis studies on3D WSN nodeslocation algorithm based on LSSVR and main work is as follows:(1)In consideration of the significant localization error of the traditional RSSI algorithm，thethesis will apply SVR in WSN location algorithm and combined RSSI location algorithm withLSSVR. Firstly, improve RSSI algorithm by the reference of distances and the received powersbetween anchor nodes. Then use the improved RSSI ranging as the input vector of LSSVR machineto build a LSSVR3D WSN nodes location algorithm based on improved RSSI.(2)Then combine centroid localization algorithm and LSSVR to build a LSSVR3D WSNnodes location algorithm based on centroid localization algorithm. By increasing the training griddensity near the unknown nodes, the training function learning process of LSSVR has beenimproved. The simulation show that in the3D environment where the nodes are randomlydistributed, the average localization error of LSSVR based on centroid localization algorithm hasgreatly reduced, compared with the traditional LSSVR location algorithm.(3)By researching and improving the parameters which is affected the RBF kernelperformance, the LSSVR regression learning machine could obtain more accurate localization．...