On concepts of performance parameters for channels

Abstract

Among the mostly investigated parameters for noisy channels are code size, error probability in decoding, block length; rate, capacity, reliability function; delay, complexity of coding. There are several statements about connections between these quantities. They carry names like "coding theorem", "converse theorem" (weak, strong, ...), "direct theorem", "capacity theorem", "lower bound", "upper bound", etc. There are analogous notions for source coding. This note has become necessary after the author noticed that Information Theory suffers from a lack of precision in terminology. Its purpose is to open a discussion about this situation with the goal to gain more clarity. There is also some confusion concerning the scopes of analytical and combinatorial methods in probabilistic coding theory, particularly in the theory of identification. We present a covering (or approximation) lemma for hypergraphs, which especially makes strong converse proofs in this area transparent and dramatically simplifies them. © Springer-Verlag Berlin Heidelberg 2006.