Analysis of RSSI-Based Localization for Room Level Deployment Accuracy

Beacon-based localization using Received Signal Strength Indicator (RSSI) is a common approach for indoor localization. However, signal variability remains a significant issue for precision localization accuracy, and existing methods involving extensive calibration and modeling to improve performance require significant deployment effort and specialization. Yet, for many applications, localization with room-level resolution is sufficient. This thesis explores configurations of a beacon-based system designed to capture receiver localization at the room level.

The accuracy of room level localization is assessed with regard to transmitter deployment schemes including generalized placement, quantity, and specific layout. Length of localization window is evaluated to ensure data quality and as proof of concept for real time operation. Six proximity localization algorithms are implemented and evaluated for their ability to accurately localize to the correct room of a floorplan.

Results indicate that configurations in which transmitters are placed in the center of room walls provide higher localization accuracy, and a minimum of 2 of RSSI beacon packets are necessary to process and stabilize optimal localization. However, analysis of algorithm performance, transmitter quantity, and layout reveals tradeoffs between accuracy, resources, and deployment effort. Configurations with 4 transmitters deployed per room coupled with a “Best of 5” voting algorithm delivered the highest accuracy of 0.9486 on average; however, maximal accuracy of algorithm performance only increases marginally as the quantity of beacons is increased, so true deployment optimum is dependent on equipment constraints and acceptable uncertainty for room-level localization accuracy.