The cliff effect is a phenomenon evidenced when the quality of received data drops abruptly when the channel quality falls below a critical point and does not improve once the channel quality surpasses this point. In modern networks (content delivery networks (CDNs), mobile, wireless), when content is transmitted over diverse channels to heterogeneous users, the cliff effect becomes a major impediment. In simultaneous video delivery to multiple users, the users with channel quality below the critical point will receive unwatchable streams, whereas those whose channel quality is well above it will not see any improvement. We propose a multiple description based joint source-channel coding approach to suppress the cliff effect in video delivery, which can be optimized according to a statistical description of the channels, and specific requirements of the application. After introducing the analytical model of the proposed approach, we describe two possible strategies to modify state-of-the-art video codecs. This involves design of a new data processing block based on linear complexity encoding by sparse codes, which is our ongoing work.
ASJC Scopus subject areas
- Electrical and Electronic Engineering