This paper challenges the conventional wisdom that video redundancy should be removed as much as possible for efficient communications. We discover that, by keeping spatial redundancy at the sender and properly utilizing it at the receiver, we can build a more robust and even more efficient wireless video communication system than existing ones.
In the proposed framework, inter-frame (temporal) redundancy in video is removed at the encoder, but intra-frame (spatial) redundancy is retained. In doing so, pixel values after a transform-domain scaling are directly transmitted with amplitude modulation. At the receiver, spatial redundancy is utilized by image denoising. Note that denoising in our decoder is not a post-processing, but have to be immediately performed on channel output. We implement the video communication system which we call Cactus on SORA platform. The image denoising processing is made real-time through GPU implementation. Cactus is extensively evaluated in 802.11a/g WLAN environment. On average, Cactus outperforms SoftCast by 4.7 dB in video PSNR and is robust to packet losses.