Error Resilient Communication and Processing Techniques for Streaming Data

In this project, I focus on low-delay error correction codes for streaming data at the application layer (c.f. Figure 1). Forward error correction codes designed for streaming sources require that (a) the channel input stream be produced sequentially from the source stream (b) the decoder sequentially reconsructs the source stream as it observes the channel output. Naturally both the optimal structure and the fundamental limits of streaming codes are expected to be different from classical error correction codes. For example, it is well known that the Shannon capacity of an erasure channel only depends on the fraction of erasures. However, when delay constraints are imposed, the actual pattern of packet losses also becomes relevant. The decoding delay over channels which introduce burst losses can be very different than over channels which only introduce isolated losses. In related works, suitable adaptations of block codes to streaming applications have been studied in many prior works. In reference [4] a class of optimal streaming codes - Maximally Short (MS) 3 Codes - are proposed for burst erasure channels. However, we consider an alternative construction which is a slight generalization of these constructions...

Ahmed Badr
Superviseur universitaire: 
Dr. Ahmed Badr
Project Year: