Hardware and software design of BMW system for multi-floor localization

Abstract

Abstract Although the Micro Electro Mechanical System (MEMS) sensors are capable of providing short-term high positioning accuracy, every positioning result significantly depends on the historical ones, which inevitably leads to the long-term error accumulation. The Bluetooth Low Energy (BLE) is independent of the accumulative error, but the positioning accuracy is suffered by the irregular jump error resulted from the Received Signal Strength Indicator (RSSI) jitter. Considering the requirement of accurate, seamless, and consecutive positioning by the existing commercial systems, we propose a new integrated BLE and MEMS Wireless (BMW) system for multi-floor positioning. In concrete terms, first of all, the way of fingerprint database construction with the reduced workload is introduced. Second, the fingerprint database is denoised by the process of affinity propagation clustering, outlier detection, and RSSI filtering. Third, the robust M estimation-based extended Kalman filter is applied to estimate the two-dimensional coordinates of the target on each floor. Finally, the barometer data are used to calculate the height of the target. The extensive experimental results show that the proposed system can not only restrain the accumulative error caused by the MEMS sensors but also eliminate the irregular jump error from the BLE RSSI jitter. In an actual multi-floor environment, the proposed system is verified to be able to achieve the Root Mean Square (RMS) positioning error within 1 m.

Keywords

• Multi-floor positioning
• Data fusion
• Robust M estimation
• BLE
• MEMS