This paper presents a technique to reduce the number of states in the trell
is of a conventional partial response Viterbi detector (PR-VD) using prelim
inary decisions. The preliminary decisions are obtained from a lower order
PR-VD which are used to shorten the tail of the higher order partial respon
se. Since it does not use decision feedback to shorten the tail of the high
er order PR, it does not suffer from error propagation. Theoretical analysi
s is carried out to show the influence of error events in the preliminary a
nd reduced state Viterbi detectors on the system performance, and to provid
e possible ways to effectively implement the scheme. Simulation results are
provided to bring out certain key features of the approach.