Improved Decoder Likelihoods for 3G Cellular Uplink over Asynchronous Multi-Path Fading Channels

3G systems, which are expected to still be widely deployed over the next decade, utilize Direct-Sequence Code-Division Multiple-Access (DS-CDMA) which suffers mainly from the Multiple Access Interference (MAI). This paper proposes improved decoder log-likelihood ratios for the practical uplink. The basic idea is to utilize knowledge about user cross-correlations and fading coefficients to obtain conditional variances for the MAI, rather than to conventionally assume it to be Gaussian with fixed variance. The conditioning may be on the users' cross-correlations and/or the channel fading coefficients and results in a time-dependent variance. We present improved log-likelihood ratios (ILLRs) for the chip synchronous case, and derive the ILLRs for the chip-asynchronous multi-path fading channel and the raised-cosine transmit filter. Our simulation framework applies system parameters from 3G standards. The improvements for the synchronous case are much larger, as expected. However, even in the practical case of cellular uplink, the performance advantage due to ILLRs is significant approaching 3 dB.