Random Access and Source-Channel Coding Error Exponents for Multiple Access Channels

Tamás Kói

Budapest University of Technology and Economics (BUTE), Hungary

Abstract:

We address a version of the random access model of J. Luo and A. Ephremides (supplemented by Z. Wang and J. Luo), which is similar to a model studied for one-way channels by I. Csiszar. The results of I. Csiszar are generalized to (discrete memoryless) multiple access channels (MACs). A two-senders random access model is introduced, in which the senders have codebook libraries with constant composition codebooks for multiple rate choices. The error exponent of Y. Liu and B.L. Hughes for an individual codebook pair is shown to be simultaneously achievable for each codebook pair in the codebook libraries, supplemented with collision detection. This is achieved via universal decoder. Moreover, achievable joint source-channel coding error exponents for transmitting independent sources over a MAC are given, admitting improvements when error free 0 rate communication is allowed between the two senders. The most direct extension of the results of I. Csiszar is obtained in the latter case. The work has been appeared in Transaction on Information Theory 2015 june.

Biography:

Tamás Kói was born in Budapest, Hungary, in 1986. He received the M.Sc. degree (with honors) in mathematics from the Budapest University of Technology and Economics (BUTE), Hungary, in 2009. He is currently working toward Ph.D. degree in mathematics at BUTE. From 2009 until 2013, he was at BUTE in various research jobs supported by the Government of Hungary. In 2013, he joined the faculty of BUTE, where he is assistant lecturer in the Department of Stochastics and part time research assistant at the MTA-BME Stochastics Research Group. His research interests include information theory, statistics and economics.