Performance of linear separation of CDMA signals with FEC coding

Many researchers have recently considered multi-user detection of signals on the multiple access channel, including the CDMA channel. The area has become highly topical because of the use of CDMA in current and proposed cellular personal communications systems, such as IS-95. Verdu (1986) introduced maximum-likelihood multi-user detection, and showed that it is able to enhance the capacity of CDMA systems. However, since it is computationally infeasible in practical situations, a number of reduced complexity techniques have been developed. One such technique (also due to Verdu) which has attracted considerable attention because of its relatively low complexity, is linear signal separation, or decorrelation. The cost of eliminating multi-user interference in linear signal separation is an enhancement of thermal noise and external interference. This increases rapidly as the number of signals to be separated approaches the number of degrees of freedom of the signals, which is equal to the number of sequence “chips” per channel symbol, after coding. Hence as code rate decreases, noise enhancement increases, setting a limit on the number of users possible and effectively preventing the use of low rate codes. In this paper we consider jointly signal separation and coding, and in particular the trade-off between the techniques: the effect of varying code rate for constant bandwidth efficiency