Machine Learning Model Compression for Efficient Indoor Localization on Embedded Platforms

Global positioning systems (GPS) have reinvented the way we localize and navigate in the outdoor environment. However, the poor reception of GPS signals indoors makes it unsuitable for indoor localization. Fingerprinting-based indoor localization is one of the most promising ways to meet this challenge. Unfortunately, most work in the domain fails to resolve challenges associated with deployment on resource-limited embedded devices. This chapter proposes a compression-aware and high-accuracy deep learning framework called CHISEL that outperforms the best-known works in the area while maintaining localization robustness on embedded platforms.