Sliding Window Integrated Weighting Algorithm for Positioning Indoor Active Objects

The modern industry necessitates gaining the real-time location of indoor objects to trigger related services and enhance efficiency. To devise effective localization algorithms concerning various metrics under different scenarios becomes much essential. Bluetooth Low Energy (BLE) comes to be a popular positioning technology, attributing to its low cost, low energy consumption, and ubiquity. In this work, the Sliding Window Integrated Mean-Max Weighting (SWIMMING) algorithm based on BLE has been proposed to locate moving objects indoors. The maximum value of signal strength within a timeslot, which attracts less attention in previous work, is regarded as one of the key features in our model. Moreover, Gaussian Processes (GPs) are used to filter raw data in advance so as to mitigate the effect of signal multipath fading. The warehouse scene is set as the background here for experiments since an actual requirement derived from the practice drives detecting the location of workers for operations advancement. BLE 5.0 is configured as the infrastructure for this test, which is new to the market and shows more stable signal transmission. A quantitative comparison with applying Kalman Filter, neural network and other approaches has been carried out. And the results demonstrate that our algorithm takes advantage of high location accuracy, cost-effective computational energy consumption, and ease of deployment. The SWIMMING algorithm achieves errors less than 1.5m 90% of the time under such dynamic conditions. Besides, other practical experiences are disclosed as a reference for future research.