In this paper we consider an additive white Gaussian noise multiple-ac
cess channel and investigate the performance degradation of several mu
ltiuser detectors due to imperfect knowledge of channel parameters, in
duced by either estimation errors or sampling/quantization of the rece
iver's sufficient statistics. The symbol error rates and the asymptoti
c multiuser efficiencies (AME's) of these detectors are examined as fu
nctions of both true and approximated channel parameters for 2 asynchr
onous users. The results quantify the loss of near-far resistance due
to parameter estimation errors and demonstrate that detectors which ar
e optimal for ideal conditions are often more sensitive to parameter m
ismatch. Examples are given which illustrate a reverse ordering of the
symbol error rates for these detectors in the high signal-to-noise re
gime, and also illustrate cases when the conventional matched-filter d
etector exhibits superior performance over multiuser detectors.