The use of multiple antennas at the transmitter and the receiver is considered for the uplink of cellular communication systems. The achievable spectral efficiency in bits/s/Hz is used as the criterion for comparing various design choices. The focus is on wideband code-division multiple-access (CDMA) systems when the receiver uses the matched-filter or the minimum mean-squared error detector, followed by single-user decoders. The spreading sequences of the CDMA system are assumed to be random across the users, but could be dependent across the transmit antennas of each user. Using analytical results in the large system asymptote, guidelines are provided for the sequence design across the transmit antennas and for choosing the number of antennas. In addition, comparisons are made between (random) CDMA and orthogonal multiaccess with multiple antennas. It is shown that CDMA, even with single-user decoding, can outperform orthogonal multiaccess when the number of receive antennas is sufficiently large.

General definitions of spreading and coding are given based on the notion of Shannon bandwidth introduced by Massey (1994), with the goal of distinguishing these operations for signaling with bandwidth redundancy. These definitions are shown to lead to a separation result: every bandwidth redundancy scheme can be expressed as a concatenation of coding followed by spreading. The coding-spreading tradeoff problem is then studied for a code division multiple access (CDMA) system in which the receiver processes the received signal by using a user-separating front-end, which feeds into autonomous single-user decoders. Under the single-user decoding setting, it is established that the linear minimum mean square error (LMMSE) front-end multiuser detector is optimum among all front-ends that are constrained to use only spreading information. Also, conditions are given for the single-user decoders to ignore spreading information without losing optimality. An example illustrating the coding-spreading tradeoff optimization for a direct sequence CDMA system with random spreading is given. Single-cell and multicell scenarios are considered in the optimization, and a comparison is made of the spectral efficiencies that can be achieved with the conventional matched filter and LMMSE front-ends.

The analysis of linear minimum mean-square error (MMSE) detection in a band-limited code-division multiple-access (CDMA) system that employs random spreading sequences is considered. The key features of the analysis are that the users are allowed to be completely asynchronous, and that the chip waveform is assumed to be the ideal Nyquist sinc function. It is shown that the asymptotic signal-to-interference ratio (SIR) at the detector output is the same as that in an equivalent chip-synchronous system. It is hence been established that synchronous analyses of linear MMSE detection can provide useful guidelines for the performance in asynchronous band-limited systems.

The problem of generating discrete sufficient statistics for signal processing in code-division multiple-access (CDMA) systems is considered in the context of underlying channel bandwidth restrictions. Discretization schemes are identified for (approximately) bandlimited CDMA systems, and a notion of approximate sufficiency is introduced. The role of chip-matched filtering in generating accurate discrete statistics is explored. The impact of approximate sufficiency on performance is studied in three cases: conventional matched filter (MF) detection, minimum mean-squared-error detection, and delay acquisition. It is shown that for waveforms limited to a chip interval, sampling the chip-MF output at the chip rate can lead to a significant degradation in performance. Then, with equal bandwidth and equal rate constraints, the performance with different chip waveforms is compared. In all three cases above, it is demonstrated that multichip waveforms that approximate Nyquist sine pulses achieve the best performance, with the commonly used rectangular chip pulse being severely inferior. However, the results also indicate that it is possible to approach the best performance with well-designed chip waveforms limited to a chip interval, as long as the chip-MF output is sampled above the Nyquist rate.

The problem of estimating the propagation delay of a new user in a coded band-limited DS/CDMA system in the presence of multiple access interference (MAI) is considered. MAI-resistant acquisition schemes are developed for a general CDMA system without the constraint that the spreading sequences of the users repeat every symbol period. It is assumed that the spreading sequences and delays of the interfering users are known. However, knowledge of their amplitudes, which would need estimation, is not assumed, and their unreliable code-symbol estimates are not used. Under this scenario, acquisition schemes are derived based on the maximum-likelihood (ML) criterion. The performance of an approximation to the ML scheme is analyzed using Gaussian approximations and by assuming that the chip boundaries of the new user are known a priori. Simulations show that the analysis is reasonably accurate for parameters in the realm of practical interest.

The scenario considered is one where a single new user is to be acquired on the reverse link by the base station, and where the channel parameters of the interfering users are known. Following a minimum mean squared error (MMSE) strategy for suppressing the multiaccess interference, the parameter estimation problem is posed in a maximum likelihood framework, To reduce complexity, the solution is implemented in two stages: first, the estimated tap delays are restricted to be at chip spacings; second, the number of taps is reduced by allowing for arbitrary spacing between them. The performance of the proposed techniques is studied through numerical simulations. It is shown that significant gains can be obtained by exploiting the structure of the interference and acquiring the channel parameters jointly.