Capacity bounds for channels with synchronization errors
Vortrag von Prof. Dr. Hugues Mercier
Datum: 22.04.13 Zeit: 13.00 - 14.00 Raum:
Synchronization problems are an integral part of technological systems operating in environments affected by uncertainties in timing, or time noise. These include data storage systems like magnetic and optical recording, semiconductor devices and integrated circuits, and synchronous digital communication networks. Time noise in- troduces insertions and deletions of symbols for the recipient, and as a result, systems corrupted by timing errors do not know their exact position when processing data. As the performance of modern communication systems improves, increasingly stringent synchronization constraints are required.Synchronization and additive noise are usually treated as different problems and overcome using different techniques. This being said, both have the same effect on communication channels, i.e., reducing their capacity. Unfortunately, channels corrupted by synchronization errors are poorly understood and many fundamental questions remain unanswered. Their capacity is unknown and seem very difficult to estimate. Furthermore, although it has early and often been conjectured that error-correcting codes capable of correcting timing errors could improve the overall performance of communication systems, they are quite challenging to design, which partly explains why we keep developing clever ways to avoid using them.
In this talk, I will present part of my work on channels corrupted by timing errors. I will discuss results on the capacity of such channels and present non-trivial capacity bounds for channels corrupted by synchronization errors and additive noise. I will also show that tools developed for channels with timing errors could also be of interest for a large number of problems that can be tackled using synchronization models, like speech recognition and watermarking, underwater communications, sequence alignment, reconciliation of mobile data, and packet-switched communication networks.