The Multisession Multilayer Broadcast Approach for Two Cooperating Receivers—How Many Sessions Are Required?

摘要

We consider the problem of transmission of a common message from a single source to two cooperating destination users, exhibiting different signal-to-noise ratios (SNR). The cooperation is performed over a noiseless link with a total capacity limit $C_{rm coop}$, which the users may utilize for conversation. In order to maximize the achievable throughput, it is suggested to employ a multilayer broadcast approach combined with multisession cooperation, such that every session another layer can be decoded. Since the cooperation link is limited, an efficient cooperation scheme is required. The relaying user performs compression which utilizes the side information of the decoder in the Wyner–Ziv (WZ) spirit. In multisession cooperation the compression is in the form of successive-refinement WZ coding. After the first user decodes a layer, the minimal required information is sent back to the second user to allow both of them to decode the message. The minimal required information is achievable with random coding using a binning strategy. While it is expected that multisession cooperation will be more efficient than a single-session cooperation we show that the maximal throughput is achieved with a single session cooperation. This is shown for the nonfading channel and for the ergodic fading case. Nonetheless, for the block fading channel it is already known that multilayering and multisession outperforms single session cooperation.