|
In wireless sensor node localization, distance measurements areoften supposed to be polluted by Gaussian noise. Throughexperiments on a practical wireless sensor network (WSN) in whichsensor nodes use received signal strength indicator (RSSI) modulesto measure distances, we show that the uncertainty of distancemeasurements can be approximated as the combination of small denseGaussian noise and large sparse errors. To handle this mixed noiseand improve the localization accuracy, we propose a novel errorcorrection algorithm for the wireless sensor node localizationproblem based on the sparse prior of large errors as well as thetechnique of semidefinite programming (SDP) relaxation. Simulationresults demonstrate that the proposed algorithm works well underthe existence of the mixed noise and outperforms several existingalgorithms. |
|
Keywords:Wireless sensor network (WSN), localization, error correction, semidefinite programming (SDP) relaxation |
|