Real-time APD-based visible light positioning system with quality-aware RSS selection and embedded tracking

This paper presents a robust received signal strength (RSS)-based visible light positioning system for embedded real-time indoor tracking. To address the sensitivity of conventional RSS-based methods to low-signal-to-noise measurements and the high computational complexity of real-time implementations, a hardware-software co-designed architecture is developed on a Xilinx Zynq-7000 system-on-chip (SoC) platform. At the hardware level, the programmable logic (PL) implements adaptive signal conditioning, frame synchronization, light-emitting diode (LED) identification, and state-based protocol detection to ensure reliable RSS acquisition under noisy conditions. On the software side, the processing system (PS) performs cascaded filtering and adaptive anchor selection to suppress unreliable measurements and mitigate bias caused by low-quality LEDs. A hierarchical search strategy further exploits inter-frame continuity to reduce computational complexity for real-time tracking. Experimental results under line-of-sight (LOS)-dominant indoor conditions demonstrate an average static positioning error of 4.98 cm and a position update rate of 15 Hz, with stable tracking across the tested motion speeds.