Abstract
This paper addresses the challenge of achieving reliable, high-speed optical camera communication (OCC) in mobile environments with multiple LED transmitters. A novel RS-OFDM-based OCC system is proposed to improve decoding accuracy, tracking robustness, and throughput under motion and interference. This solution targets applications such as IoT, ITS, and e-health, where the RF spectrum is limited. While rolling-shutter OFDM (RS-OFDM) improves spectral efficiency, it still struggles in mobility scenarios with multiple transmitters. The proposed system incorporates a multiprocessing architecture, YOLOv11n, for real-time LED detection, an advanced ByteTrack with re-identification (ReID), and an extended Kalman filter (EKF) for reliable tracking. An adaptive region of interest (RoI) improves decoding efficiency, while the denoising autoencoder (DAE-RS) mitigates ambient noise. Parallel decoding enables simultaneous message extraction from multiple transmitters within a single frame. Message sequencing and buffer management ensure data integrity through track ID and sequence number alignment. Experiments show high tracking accuracy (MOTA: 0.9861, MOTP: 0.1075), zero ID switches, and effective noise reduction (RMSE: 65.2979). YOLOv11n provides fast inference (2.4 ms) and high detection accuracy (mAP@50: 0.989). The frame rate increases from 11.30 to 16.83 frames per second. The system maintains 5.03 kbps (single) and 4.16 kbps per LED (dual) at 10−4 BER over 4 meters.
Funding Information
  • National Research Foundation of Korea
  • Korean government (2022R1A2C1007884)