The Indoor Wireless Localization Algorithm Research Based On Markov Random Field

In recent years, the Wireless Sensor Networks based on ZigBee technology developsrapidly, compare to other wireless communications technologies, it has a low-cost, lowpower, high reliability and networking flexible advantages. Indoor wireless localization isan important application of Wireless Sensor Networks based on ZigBee technoloy, itmakes up for the defect of the GPS can not be achieved in the indoor localization and hasimportant application value in the airport hall, supermarket, underground car parks,exhibition halls, mines and other environments. Common localization method mainlyinclude: Time of Arrival (TOA), Time Difference of Arrival (TDOA), Angle of Arrival(AOA), Received Signal Strength Indicator (RSSI). TOA, TDOA, AOA methods havehigh requirements for hardware, so use of low-cost RSSI positioning is an importantresearch part of indoor wireless localization, but RSSI is susceptible to attenuation,interference and other factors, now there is no accurate localization method.The existing localization algorithms based on RSSI commonly use methods ofprobability, they process equally all the RSSI values collected by the reference nodes, andhave no ability to identify abnormal RSSI values which deviate from the normal RSSIvalues largely, result in the localization precision is low. Therefore, propose a localizationalgorithm based on hierarchical Markov Random Field model, the localization area isdivided into a number of size equal small areas, use dependencies relationship of adjacentreference nodes as filter means, to remove the abnormal RSSI values by the assumption ofa priori probability, in the formed random field, maximum a posterior probability, usingthe belief propagation algorithm to get the estimated area of target node, and then dividethis area into some smaller areas again, such hierarchical computing continues, finally, thegeometric center of the area is the estimated position of target node.The proposed algorithm is verified through simulation and measured data, thesimulation data generated by Matlab, the measured data collected by the reference nodesin Wireless Sensor Networks using ZigBee technology. Compared with the previousprobabilistic algorithms, the proposed algorithm has better performance in accuracy and stability, through the chart shows the simulation data with different parameters on theimpact of the positioning results, and analyzes the reasons.