In this paper, we propose a channel classification method that identifies t
he delay path profile of Rayleigh fading channels, which can be used in con
junction with conventional maximum-likelihood sequence estimation (MLSE) re
ceivers. The proposed method determines the appropriate number of delay tap
s for the MLSE trellis, based on a decision variable obtained from multiple
traffic bursts. By formulating the decision variable using the F-distribut
ion, we derive exact expressions for misclassification probability. A new M
LSE receiver structure that utilizes the channel classification method is d
escribed, and bit error rate (BER) simulation results are presented to demo
nstrate the performance improvement. The particular application which is di
scussed is the IS-136 TDMA standard.