The question which of the access methods is best does not have a single answer. Based on the discussion above FDMA is only suited for applications like cordless telephone with very small cells and sub-microsecond delay spreads. In cellular systems and for most visions of personal communication systems the choice reduces to TDMA versus CDMA.
In terms of complexity, TDMA receivers require adaptive, nonlinear equalizers when operating in environments with large delay spreads. CDMA systems, in turn, need RAKE receivers and sophisticated power control algorithms. In the future, some form of multiple-access interference rejection is likely to be implemented as well. Time synchronization is required in both systems albeit for different reasons. The additional complexity for coding and interleaving is comparable for both access methods.
An often quoted advantage of CDMA systems is the fact that the performance will degrade gracefully as the load increases. In TDMA systems, in turn, requests will have to be blocked once all channels in a cell are in use. Hence, there is a hard limit on the number of channels per cell. However, there are proposals for extended TDMA systems which incorporate re-assignment of channels during speech pauses. Not only would such extended TDMA systems match the advantage of CDMA systems from the exploitation of speech pauses but they would also lead to a soft limit on the system capacity. The extended TDMA proposals would implement the statistical multiplexing of the user data by means of the Packet Reservation Multiple Access protocol [Goodman, 1991a]. The increase in capacity depends on the acceptable packet loss rate; in other words, small increases in the load lead to small increases in the packet loss probability.
Many comparisons in terms of capacity between TDMA and CDMA can be found in the recent literature. However, such comparisons are often invalidated by making assumptions which favor one access method over the other. An important exception constitutes the recent paper by Wyner [Wyner, 1994]. Under a simplified model which still captures the essence of cellular systems he computes the Shannon capacity. Highlights of his results include:
More research along this avenue is necessary to confirm the validity of the results. In particular, incorporation of realistic channel models into the analysis is required. However, this work represents a substantial step towards quantifying capacity increases achievable with CDMA.
