Using Pixel-per-bit Neural Network for Two Rolling Shutter Patterns Decoding in Optical Camera Communication (OCC)

Optical wireless communication (OWC) has received increasing attention recently. One application of OWC uses image sensor as optical receiver, which is known as optical camera communication (OCC). Rolling shutter mode in OCC allows data rate higher than the camera frame-rate operation. However, the row by row exposure delay will create decoding challenge, and high inter-symbol interference (ISI) can be observed in the rolling shutter pattern. In this work, we propose and demonstrate an OCC system employing pixel-per-bit (PPB) as label for the deep neural network (PPB-NN). Besides, two rolling shutter patterns emitted by two LED light panels can be decoded simultaneously. The proposed PPB-NN is also compared with other decoding schemes in the literature, and the experimental results revealed that the proposed scheme can achieve a better bit-error ratio (BER) performance.