A weakly secure multiple description coding scheme in lossy multipath channels for fine-grained SVC streaming
- PDF / 1,635,058 Bytes
- 15 Pages / 595.276 x 790.866 pts Page_size
- 89 Downloads / 195 Views
REGULAR PAPER
A weakly secure multiple description coding scheme in lossy multipath channels for fine‑grained SVC streaming Qiushi Gong1 Received: 22 September 2019 / Accepted: 8 July 2020 © China Computer Federation (CCF) 2020
Abstract Video streaming service is growing rapidly these years. The quality of video streaming may suffer from issues like packet loss and solutions like multiple description coding (MDC) has been shown to provide flexible and distortion-rate optimal video streaming transmission over packet-loss links. On the other hand, the contents of stream are prone to attacks like eavesdropping while MDC is not specifically designed for security purpose. In this paper, fine grain scalably coded video data is channel-coded through multiple description-forward error correction (MD-FEC) coding. The resulting descriptions (packets) are then sent to a variety of channels, which may be peeked by an eavesdropper with certain probability. We adopt the weakly secure design here to reduce the quality of video obtained by the eavesdropper. Two weak security criteria for this problem are proposed, namely the deterministic overhearing peak-signal-to-noise-ratio (PSNR) constraint and the average overhearing PSNR constraint. The corresponding optimization algorithms are provided. Numerical results show that by applying weakly secure design, the PSNR value on the eavesdropper’s side will be reduced to the level at which useful details of the clip can hardly be exposed at the cost of reducing around 1–2 dB at the authorized receiver’s side. Keywords Multiple description · video streaming · Weakly secure · Fine-grained SVC
1 Introduction Video streaming service is growing rapidly and becoming one of the most popular applications on the Internet. As predicted by the Cisco white paper, video could become 82 percent of all user traffic in 2022 (Cisco 2017) which is 4 times the amount five years ago. At the same time, the coming unification of live, near-to-live, and video-on-demand via intelligent or dynamic caching will require robust solutions to replace present adaptive unicast solutions, e.g. ABR, MPEG-DASH, etc. now dominant in video-on-demand (VoD) streams. The work of multiple description coding with forward error correction (MD-FEC) for video (Puri and Ramchandran 1999; Mohr et al. 2000; Albanese et al. 1996) ensures effective video unicast over typical network environments. Both stream flexibility and distortion-rate optimality can be realized through MD-FEC refreshed on a Group-of-Pictures * Qiushi Gong [email protected] 1
School of Cyber Science and Engineering, Southeast University, Nanjing, China
(GOP) basis. Under MD-FEC, chunks of fine grain scalably compressed video are coded as multiple descriptions (packets1) and protected from channel loss via maximum distance separable (MDS) coding such as Reed–Solomon (RS). The average received distortion is then just a function of the number of packets received. In fact, MD-FEC can be seen as applying unequal cross-packet FEC to an existing highly scalable
Data Loading...
