The number of transmit antennas that can be employed in the context of least-squares (LS) channel estimation contrived for orthogonal frequency division multiplexing (OFDM) systems employing multiple transmit antennas is limited by the ratio of the number of subcarriers and the number of significant channel impulse response (CIR)-related taps. In order to allow for more complex scenarios in terms of the number of transmit antennas and users supported, CIR-related tap prediction-filtering-based parallel interference cancellation (PIC)-assisted decision-directed channel estimation (DDCE) is investigated. New explicit expressions are derived for the estimator's mean-square error (MSE), and a new iterative procedure is devised for the offline optimization of the CIR-related tap predictor coefficients. These new expressions are capable of accounting for the estimator's novel recursive structure. In the context of our performance results, it is demonstrated, for example, that the estimator is capable of supporting L=16 transmit antennas, when assuming K=512 subcarriers and K/sub 0/=64 significant CIR taps, while LS-optimized DDCE would be limited to employing L=8 transmit antennas.

OFDM systems employing multiple transmit antennas have recently drawn wide interest in the context of both space-time coded- and multi-user space-division multiple access (SDMA) arrangements. A prerequisite for using coherent detection at the receiver is the availability of reliable channel transfer factor estimates. Robust parallel interference cancellation (PIC) assisted decision-directed channel estimation (DDCE) has been shown in the literature to be also applicable to scenarios, where the number of users is in excess of the number of OFDM subcarriers - normalized to the number of channel impulse response (CIR) related taps to be estimated - which imposed a limitation in the context of least-squares assisted DDCE techniques invoked in conjunction with multiple transmit antennas. In this paper we will demonstrate that the recursive least-squares (RLS) algorithm is applicable to optimizing the predictors' coefficients on CIR-related tap-by-tap basis. Compared to 'robust', non-adaptive approaches the proposed solution has the advantage of a potentially lower estimation MSE and a higher resilience to erroneous subcarrier symbol decisions.

Subband adaptive orthogonal frequency division multiplexing (OFDM) constitutes an effective method of compensating for the frequency selective fading incurred by mobile receivers. The channel transfer function estimation error increases both as a function of the Doppler frequency and that of the subband modem mode signalling delay. This leads to a degradation of the system's performance. In this article a 2D-MMSE channel-predictor is investigated as a means of compensating for these effects, which incorporates the scheme proposed Tufvesson, Faulkner and Maseng, (see IEEE Proc. of 6th International Workshop on Intelligent Signal Processing and Communications Systems, Melbourne, Australia, p.57-33, November 1998) into an adaptive OFDM scheme.

Multiuser detection assisted, multiple transmit antenna based OFDM arrangements are studied in the context of HIPERLAN 2-like systems. It is demonstrated that the system user capacity can be improved with the aid of unique spatial user signatures, hence supporting a multiplicity of users. The maximum likelihood sequence estimation (MLSE) detection algorithm outperformed the minimum mean square error (MMSE) scheme by about 5 dB in terms of the required signal-to-noise ratio (SNR) and this performance gain manifested itself also in terms of the system improved video performance.

Multiple reception antenna assisted space division multiple access (SDMA) techniques designed for wireless OFDM systems have recently drawn wide interest. These techniques facilitate the implementation of effective multiuser detection algorithms at the receiver, such as minimum mean-square error (MMSE) detection, successive interference cancellation (SIC), parallel interference cancellation (PIC) or maximum likelihood (ML) detection. We investigate the performance of the MMSE and PIC detection techniques in the context of a Walsh-Hadamard spreading assisted SDMA OFDM system.

Turbo-coded burst-by-burst adaptive orthogonal frequency division multiplex (AOFDM) wide-band speech transceivers are proposed. A constant throughput adaptive OFDM transceiver was designed and benchmarked against a time-variant rate scheme. The proposed joint adaptation of source-codec, channel-codec, and modulation regime results in attractive, robust, high-quality audio systems, capable of conveying near-unimpaired wide-band audio signals over fading dispersive channels for signal-to-noise ratios (SNR) in excess of about 5 dB.