A scalable Wi-Fi based localization approach

Abstract

This paper proposes a scalable method which allows deploying the fingerprint Wi-Fi localization algorithm for different devices. The original fingerprint localization algorithm performs accurately only if the device used in the testing phase is the same as the device used in the training phase. When a different device is used in the testing phase, a time-consuming re-training step (in the order of hours or days) is required to achieve the equivalent degree of accuracy. Our proposed approach replaces the re-training step with a short period of calibration (in the order of a few minutes), which can be done transparently to the user. To validate our approach, we collected data from a large scale experiment (14 laptops and 2 smartphones with 224-hour of collected data) to evaluate the performance.