|
Title of the Proposal: Multipath Streaming Codes Enabling reliable low-latency communication is a key target of 5G cellular systems and streaming codes represent a promising means of enabling this goal to be achieved. By a streaming code, we mean a forward error-correcting (FEC) scheme that operates on a stream of packets, and that enables efficient recovery from dropped or erased packets under a strict decoding-delay constraint. Over the past few years, our research group has been successful in constructing a class of streaming codes that may be regarded as convolutional codes at the packet level, but are yet built out of scalar block codes. Our past research assumed a single path from source to destination and a Gilbert-Elliott model for the channel. Accordingly, the codes constructed were designed to handle equally effectively, either a burst of erasures or else, a set of random erasures. During this period, we have actively sought feedback from industry and this proposal aims to broaden the scope of streaming codes by taking into account the received feedback. From the feedback received, we gathered that augmented or virtual reality and intelligent vehicle transportation represent two of the more pressing applications that streaming codes might wish to target. It was also pointed out to us that deep fades over a wireless link represent a particularly serious challenge in ensuring reliability with low latency. The normal approach to overcoming deep fades in wireless communication is to use diversity and so this suggested to us that we use the natural diversity available in a network, namely path diversity. In 5G setups such as integrated access backhaul (IAB) and dual connectivity (DC), there are known to exist, multiple, link-disjoint paths from source to destination. This proposal aims at extending our past research on streaming codes with the target of bringing our research another step closer to applications, by using path diversity to overcome the challenges posed by a deep fade. Our goal is to understand the fundamental trade-off between reliability, latency and rate for multipath case and provide efficient multipath streaming code constructions. |