The performance of maximal ratio combiner (MRC) diversity reception wi
th L correlated branches in Nakagami fading is presented. Bit error ra
te (BER) is evaluated for binary signals starting from the L-variate m
oment-generating function (MGF) of the random input power vector. In t
he case of Nakagami fading the L-variate statistics are not uniquely d
etermined by marginal distributions and covariance matrix; in particul
ar, two different joint MGFs are presented for the L-variate Nakagami
vector, which give the same marginal density functions and covariance
matrix and are easily generated. The two MGFs yield two different clos
ed form expressions for the BER of the MRC diversity with generic bran
ch correlation. The sensible difference in performances suggests that
marginal probability density functions (pdfs) and power covariance mat
rix are not sufficient to completely characterize correlated Nakagami
fading. (C) 1998 Elsevier Science B.V. All rights reserved.