Multistatic Narrowband Localization in Backscatter Sensor Networks

Recent developments in signal processing for backscatter (or simply scatter) ultra-low power radio sensor networks have dramatically extended their range and coverage; thus, estimating the location of each tag becomes both valuable and challenging. In multistatic architectures, each scatter radio tag reflects the illuminating signals, transmitted from multiple low-cost emitters. The received signal strength (RSS) measurements at a single low-cost reader are used as input in custom particle filtering (PF) algorithms, for tag location estimation. Free-space and two-ray wireless propagation large-scale path-loss models are considered, alongside a small-scale fading model, different PF initialization and resampling techniques and a directional reader antenna model. Extensive simulations for a 100 m x 100 m grid offered root mean squared localization error (RMSE) as low as 1.6 m, with specific PF resampling techniques and exploiting reader antenna directionality. Outdoor experimental results in a 27 m x 32 m grid offered 1.5 m RMSE.