The Queued-Code in Finite-State Markov Fading Channels with Large Delay Bounds

A 'queued-code' is a novel channel code, whose input rate is instantaneously adapted to the channel state information, assumed to be known to the transmitter. Previously, a queued-code was demonstrated for i.i.d. block fading channels, demonstrating significant gains, in terms of error exponent, over a fixed rate code. This paper demonstrates the performance gain of the queued-code in the more realistic case of a finite-state Markov fading channel. Similar to the block-fading result, the analysis required combines the concept of random coding bounds from information theory, and the concept of effective capacity, arising from large deviation theory analysis of queues. However, a result on the limiting moment generating function of Markov chains is the additional tool required for the analysis