Joint source/channel coding of WWAN multicast video for a cooperative peer-to-peer collective using structured network coding

Because of frequent wireless packet losses and inapplicability of retransmission-based schemes due to the wellknown NAK implosion problem, providing high quality video multicast over wireless wide area networks (WWAN) remains difficult. Traditional joint source/channel coding schemes for video multicast-optimal bit allocation among source coding and channel coding such as forward error correction (FEC) subject to a bitrate constraint-target a chosen nth-percentile WWAN user. Not only is FEC bitwise expensive, users with poorer reception than nth-percentile user suffer substantial channel losses, while users with better reception have more channel coding than necessary, meaning too few bits are devoted for source coding to reduce quantization noise and sub-optimal video quality. Instead, in this paper we perform joint source/channel coding of WWAN video multicast for an entire collective of multi-homed ad-hoc peers in the same multicast group and connected via wireless local area networks (WLAN). In a cooperative peer-to-peer repair (CPR) scenario, after each peer received a different subset of WWAN packets, the peer group repairs WWAN losses locally by packet-forwarding to each other via WLAN. From an end-to-end system view, CPR means that a packet can be transmitted from source to a peer either via WWAN directly, or via WLAN local repairs exploiting neighboring peers' WWAN links; the overall more general transmission condition means a clever joint source/channel coding scheme can now allocate more bits to source coding without suffering more packet losses, leading to higher video quality. To efficiently implement both WWAN FEC and WLAN CPR repairs, we propose to use network coding for this dual purpose to reduce decoding complexity at the peers. We show through simulations that using our proposed scheme dramatically improves video quality over existing optimization scheme where joint source/channel coding was performed, but WLAN CPR was not used, by up to 8.4 dB, and over scheme when WLAN CPR and WWAN joint source/channel coding were performed separately by up to 4.4 dB.