Fixed-Memory Capacity Bounds for the Gilbert-Elliott Channel
Yutong Han, Technical University of Munich, Germany; Albert Guillén i Fàbregas, University of Cambridge, Universitat Pompeu Fabra, United Kingdom
Session:
Channel Capacity
Track:
9: Shannon Theory
Location:
Omega
Presentation Time:
Mon, 8 Jul, 10:45 - 11:05
Session Chair:
Amos Lapidoth, ETH Zurich
Abstract
We derive finite-memory upper and lower bounds to the entropy rate of binary 2-state hidden Markov models. These directly provide upper and lower bounds to the capacity of the Gilbert-Elliott channel. As the memory increases, the bounds approach the capacity of the channel. Even a simple memory-1 upper bound significantly improves over the current best upper bound by Mushkin and Bar-David.