Adaptive tracking and recording method for dynamic real-time WiFi fingerprint positioning

摘要

The overall performance of fingerprint positioning system will be improved in this paper with the adoption of technology in artificial intelligence field and optimization technology. First, at offline stage, adaptive tracking and recording method for dynamic real-time WiFi fingerprint positioning has been proposed. Regional propagation model (RPM) of indoor wireless signal has been proposed through the RSS value clustering fading characteristics observed based indoor zoning principle in clustering channel modeling algorithm. Take samples from the reference points of sparse density distribution with adoption of affinity propagation clustering technology to attain fingerprints, and then divide the indoor area into sub regions with specific number, make use of the sampling data of each sub region to set up proposed RPM path loss propagation model and predict the fingerprints of other unmeasured reference points through this model to reconstruct complete fingerprint database. Experimental testing results have shown that the RSS prediction accuracy of proposed path loss propagation model is higher than existing propagation model; at the same time, when decrease fingerprint acquisition workload above 28%, the proposed algorithm can still attain high positioning accuracy.