Expurgated Random-Coding Ensembles: Exponents, Refinements and Connections

Abstract—This paper studies expurgated random-coding bounds and exponents for channels with maximum-metric de-coding. A simple non-asymptotic bound is shown to attain an exponent which coincides with that of Csiszár and Körner for discrete memoryless channels, while remaining valid for continuous alphabets. Using an alternative approach based on statistical-mechanical methods, an exponent for more general channels and decoding metrics is given. I