Abstract

Information theoretically, frame synchronisation is not an issue at all. In coded communications, I will explain how the decoder can be used to achieve practically perfect frame synchronisation at no extra cost. Hence, our need to precede data blocks with a synchronisation prefix is a pure waste of resources that we are constrained to for practical complexity reasons rater than fundamental information theoretic reasons.

I will first explore how long such a prefix needs to be and explain why no rate loss is in fact incurred for synchronisation, despite the wasted prefix, depending on how noisy the channel is. Furthermore, I will discuss a quick look-in approach to frame synchronisation via the decoder where synchronisation can be achieved at lower complexity based on the decoder, without a known prefix. I will also show that this method does not work well for linear codes and introduce mixed linear/non-linear codes for which quick look-in synchronisation works as well as prefix-based synchronisation, and compare the decoding performance of the two systems for reasonable practical block lengths.

This subject of this talk is currently the object of a 4th year project under my supervision and much of the material I present results from my collaboration with a IIB student Joned Sarwar. If you miss or misunderstand this talk you’ll have a chance to catch up on the subject during his project presentation on the 3 June.