Causal Transmission of Colored Source Frames over a Packet Erasure Channel

Abstract

We propose a linear predictive quantization system for causally transmitting parallel sources with temporal memory (colored frames) over an erasure channel. By optimizing within this structure, we derive an achievability result in the high-rate limit and compare it to an upper bound on performance. The proposed system subsumes the well-known PCM and DPCM systems as special cases. While typically DPCM performs well without erasures and PCM suffers less with many erasures, we show that the proposed solution improves performance over both under all severities of erasures, with unbounded improvement in some cases.

Zero-Delay Sequential Transmission of Markov Sources Over Burst Erasure Channels

“…However this setup does not capture the effect of packet losses over a channel, where the destination has access to all the non erased symbols. To our knowledge, only reference [3] considers the setting of sequential coding over a random packet erasure channel. The source is assumed to be Gaussian, spatially i.i.d.…”

“…A class of linear predictive coding schemes is studied and an optimal scheme within this class, with respect to the excess distortion ratio metric is proposed. Our proposed coding scheme is qualitatively different from [3], [13] and involves a random binning based approach, which is inherently robust to the side-information at the decoder.…”

“…whereŝ n i denotes the reconstruction sequence and j denotes the time at which burst erasure starts in (3). The destination is not required to produce the source vectors that appear either during the burst erasure or in the period of length W following it.…”

“…The predictively-coded frames are used to improve the compression efficiency whereas Manuscript the intra-coded frames limit the amount of error propagation. Other techniques including forward error correction codes [2], leaky DPCM [3] and distributed video coding [4] can also be used to trade off the transmission rate with error propagation. Despite this, such a tradeoff is not well understood even in the case of a single isolated packet loss [5].…”

Sequential coding of Gauss-Markov sources with packet erasures and feedback

“…For random independent identically distributed (i.i.d.) packet erasures, a hybridation between pulse-code modulation (PCM) and differential PCM (DPCM), termed leaky DPCM, was proposed in [9] and analyzed for the case of very low erasure probability in [10]. The scenario in which the erasures occur in bursts was considered in [11], [12].…”