A unified approach to multirate detection for cellular systems on fast-fading channels

Nearly all multirate detection schemes proposed for IS-95 and CDMA2000 rely heavily on the assumption of an additive white Gaussian noise (AWGN) channel model. The idea is to exploit the property of symbol repetition and to test, in one way or another, the equality of the mean value of the received samples within an appropriate data block. In a fast fading environment as encountered in cellular mobile systems, however, the time-varying channel gain destroys the block-wise constant-mean structure in the received process lending the use of an AWGN channel model no longer justified. It is the correlation structure, rather than the mean value, of the received process that carries the information for multirate detection. To fully reflect this nature, we formulate the problem of multirate detection in a completely new framework ending up with a simple maximum-likelihood-based algorithm to examine the canonical correlation of the received sequence. The new test possesses the property of constant false-alarm rate thereby allowing for thresholds to be analytically determined. Computer results are also presented to demonstrate the superior performance of the new algorithm in Rayleigh fading.