Visibile light communication system with mobile phone

Abstract

Visible light communication (VLC) system is getting popular as it does not required radio connection and it can enable secure and interference-free wireless link. This make VLC an attractive solution for short-range wireless communication. This report will discuss how VLC works on screen-camera link. Previous works regarding VLC required large LCD monitors and high-speed digital cameras using complex communication and computer vision techniques. COBRA is a VLC over screen-camera links application for smartphones. It is a specially designed 2D color barcodes which is able to store a high capacity of data and achieve fast decoding speed. The objective of COBRA is to improve the streaming between small-size screen and smartphones which has a low-speed camera. COBRA is studied in this report as it was the first application that was built to support off-the-shelf smartphone platforms for VLC. It was implemented using the HSV color space for its modulation. It used a novel technique which allow code blocks to be quickly localized and extracted from low-quality image which is very common in mobile environment. COBRA was originally implemented using HSV color space for its demodulation. With this implementation, identifying a wrong color block is possible. Thus, COBRA using RGB color space was implemented and the results show a slight improvement. The main contribution of this report was to improve the performance of COBRA by using different color spaces. By using different color spaces, the modulation and demodulation were changed and a color palette was introduced to normalize the YUV value to 0 and 1 when COBRA was implemented using YUV color space. Under modulation, the pixel to be displayed is based on the YUV value which had undergo a conversion to RGB. As the original displayed pixel value is based on YUV, during demodulation, the value of each pixel is a RGB value which is required to be converted to YUV. With the converted value, it will decide color of the pixel which will be used to determine the bits of that particular block. The results shows that there is about 16.2% increase of goodput among the 5 different block size that was tested. Based on the results of COBRA with different color spaces, we are unable to achieve 100% decoding rate. Thus, we are unable to regenerate the encoded image. But with the help of Reed-Solomon forward error correction, we are able to regenerate the encoded image even though the decoding rate is below 100%.