This paper presents analytical and numerical results on the sample siz
e required to achieve a specified root mean square (rms) error in esti
mating en oi-rate for flat fading channels having complex Gaussian sta
tistics. The analysis shows that Sor the large sample sizes normally u
sed in estimating en oi rates, k, the required sample size normalized
to the required sample size Soi independent symbol fading, can be expr
essed in the form k = 1 + d beta where d is the symbol rate normalized
to the Doppler spl-ead of the channel. For a given modem, beta is a f
unction of the error probability and the order of diversity. It is sho
wn that if the Doppler spread measure used is proportional to the rms
Doppler spread, beta will be relatively insensitive ro the shape of th
e Doppler power spectrum.(1) Numerical results are presented for Lth o
rder diversity: reception of binary phase shift keying (PSK), differen
tial PSK, and frequently shift keying (FSK) signals and for five diffe
rent Doppler power spectra Ideal maximal ratio combining is assumed So
l the PSK modem, and square law; combining is assumed for the DPSK aid
FSK modems.