A comparison of modulation schemes in bandlimited AWGN channels

Abstract

In recent years, as data rates rise for seemingly decreasing available bandwidths, a great deal of research has been directed toward finding bandwidth efficient modulation schemes. Two such methods are Partial-Response Signaling and Trellis-Coded Modulation. Both of which promise performance gains in a bandlimited channel when compared to uncoded systems. This paper will compare the performance of these schemes, when applied to a QPSK system over various channel bandwidths. IntroductionAs data rate demands increase, the need for bandwidth efficient modulation schemes has driven a great deal of research. As a result, a number of complex systems have been developed. Partial-Response Signaling (PRS) [I ,2,3] and Trellis-Coded Modulation (TCM) [4,5] are two methods promising increased performance over bandlimited channels, thus allowing higher data rates through the channel or better performance in the same channel. This paper will investigate the performance of relatively simple implementations of each of these techniques -a QPSK PRS and an 8PSK TCM systemin channels of different bandwidths. By doing so, it will be determined whether or not these implementations live up to their promised performance gains.Bit error rates are estimated through baseband Monte Carlo simulation using the Signal Processing Worksystem@ (SPWTM) software package. Each system compared transmits at a rate of two information bits per channel symbol. Channel bandwidth is specified by a 6th order Butterworth filter without phase-equalization. Simulations are performed with no bandlimiting and at bandwidths equal to 3, 1.5 and 1.0 times the symbol rate, generating varying degrees of ISI.Both modulation schemes transmit symbols representing two data bits. Therefore, their performance will be judged for equal symbol rates -equal information rates -over channels of equal bandwidth. For comparison purposes, the performance of an uncoded