Training codes are introduced for the multiple-antenna, noncoherent, multiple block-Rayleigh-fading channel in which the fading coefficients, which are constant over a fixed number of dimensions (coherence interval) for each block and then change independently to a new realization, are known neither at the transmitter nor at the receiver. Each codeword of a training code consists of a part known to the receiver-used to form a minimum mean-squared error (MMSE) estimate of the channel-and a part that contains codeword(s) of a space-time block or trellis code designed for the coherent channel (in which the receiver has perfect knowledge of the channel). The channel estimate is used as if it were error-free for decoding the information-bearing part of the training codeword. Training codes are hence easily designed to have high rate and low decoding complexity by choosing the underlying coherent code to have high rate and to be efficiently decodable. Conditions for which the estimator-detector (E-D) receiver is equivalent to the optimal noncoherent receiver are established. A key performance analysis result of this paper is that the training codes when decoded with the E-D receiver achieve a diversity order of the error probability that is equal to the diversity order of the underlying coherent code. In some cases, the performance of training codes can be measured relative to coherent reception via "training efficiency," which is then optimized over the energy allocation between the training and data phases. In the limit of increasing block lengths, training codes always achieve the performance of coherent reception. The examples of training codes provided in this work have polynomial complexity in rate but an error rate comparable to the best performing unitary designs available, even though the latter require exponential decoding complexity.

The jointly optimum receiver is obtained for multi- user communications in a frequency nonselective Rayleigh-fading channel with N/sub T/ transmit antennas per user and N/sub R/ receive antennas. Based on a general analysis of quadratic receivers in zero-mean complex Gaussian vectors, asymptotically tight expressions (for high signal-to-noise ratio (SNR)) for the pairwise error probabilities are derived. Subsequently, it is shown that N/sub T/-dimensional single-user signaling suffices to provide full diversity order N=N/sub T/N/sub R/ for all the users. In other words, the presence of other users does not increase the minimum dimension required beyond what is needed for the single-user space-time channel. For the special case of low-rank "code-division multiple-access (CDMA)" signaling with N/sub T/=1 and provided the signatures of any two users are linearly independent, it is shown that the error probability of a K-user system asymptotically approaches single-user-like performance for every user. Remarkably, therefore, an increase in the number of users, and hence an increase in the aggregate spectral efficiency, does not require the users to transmit with more power to achieve single-user-like performance asymptotically. A signal design algorithm is proposed to illustrate this point and examples are given. These results are then generalized to the multiple transmit antenna case. A new (N/sub T/+1)-dimensional signaling strategy is proposed for the multiuser channel that leverages existing single-user space-time signal designs while ensuring full diversity order and single-user-like performance asymptotically for every user.

The jointly optimum multiuser noncoherent detector for differential phase-shift keying (DPSK) modulation over the generalized diversity Rayleigh-fading (GDRF) channel is derived and analyzed. The GDRF channel includes time/frequency/receiver antenna diversity and allows fading correlations between the various diversity branches of each user. Noncoherent detection here refers to the case where the receiver has neither knowledge of the instantaneous phases nor of the envelopes of the users' channels. Upper and lower bounds on the bit-error probability of the optimum detector are derived for a given user. For fast fading, when the fading coefficients vary from one symbol interval to the next (but are still essentially constant over one symbol interval), the detector asymptotically (for high signal-to-noise ratios (SNRs)) reaches an error floor, which is bounded from below and above for different fast fading scenarios. For slow fading, when the channel is constant for at least two consecutive symbol intervals, the upper bound is shown to converge asymptotically to the lower bound. Thus, the asymptotic efficiency of optimum multiuser DPSK detection can be determined and is found to be positive. In contrast to coherent detection, however, it is smaller than unity in general. Since the asymptotic efficiency is independent of the interfering users' signal strengths, the optimum detector is near-far resistant. While optimum multiuser detection is exponentially complex in the number of users, its performance provides the benchmark for suboptimal detectors. In particular, it is seen that the previously suggested post-decorrelative detectors can be far from satisfactory.

We consider the problem of designing signal constellations for the multiple transmit-multiple receive antenna Rayleigh-fading communication channel, when neither the transmitter nor the receiver know the fading. In particular, by employing the asymptotic union bound (AUB) on the probability of error, we give a new formulation of the problem of signal design for the noncoherent fading channel. Since unitary signals are optimal for this channel (in the limit of large signal-to-noise ratios SNRs), the problem can be posed in terms of packings on the Grassmanian manifold. A key difference in our approach from that of other authors is that we use a notion of distance on this manifold that is suggested by the union bound. As a consequence of our use of this distance measure, we obtain signal designs that are guaranteed to achieve the full diversity order of the channel, a result that does not hold when the chordal distance is used. We introduce a new method of recursively designing signals, termed successive updates, to approximately optimize this performance measure. We then examine the use of our signals with several convolutional codes over the fading channel. An upper bound on the bit error probability of the maximum-likelihood decoder is presented together with an asymptotic analysis of that bound.

In a previous paper we proposed a multiuser space-time modulation scheme that leverages single-user space-time constellations and guarantees both, a full diversity order, as well as an asymptotic (high SNR) single-user like performance for every user. This is achieved by multiplying each user's space-time information matrix symbol by a low-dimensional "spreading matrix." For instance, for N/sub T/-transmit antennas per user and if the single-user space-time constellation employed requires only the minimum dimension N/sub T/, no more than N/sub T/+1 dimensions are required for the common signal space of all users, i.e., each user's spreading matrix is of size (N/sub T/+1)/spl times/N/sub T/, independent of the number of users. In this paper, we present a simplified design criterion to obtain these spreading matrices by numerical optimization. Since the columns of each user's spreading matrix are constrained to be orthonormal, we propose to perform the optimization using a parameterization of the Grassmann manifold. A special feature of the simplified criterion, and thus of the resulting spreading matrices, is that they are independent of the particular single-user space-time constellations (of a given dimension), so that different spectral efficiencies can be attained without changing or redesigning the spreading matrices.

A general, asymptotic (high signal-to-noise (SNR)) error analysis is introduced for quadratic receivers in frequency-flat and multipath Rayleigh-fading channels with multiple transmit and receive antennas. Asymptotically tight expressions for the pairwise error probabilities are obtained for coherent, noncoherent, and differentially coherent space-time receivers. Not only is our unified analysis applicable to more general modulation schemes and/or channel models than previously considered, but it also reveals a hitherto unrecognized eigenvalue structure that is common to all of these problems. In addition to providing an easy recipe for computing the asymptotic pairwise error rates, we make some conclusions regarding criteria for the design of signal constellations and codes such as (a) the same design criteria apply for both correlated and independent and identically distributed (i.i.d.) fading processes and (b) for noncoherent communications, unitary signals are optimal in the sense that they minimize the asymptotic union bound.

The jointly optimum noncoherent multiuser detector for DPSK modulation in generalized diversity Rayleigh fading channels (GDRF) was presented previously (Varanasi and Brehler, 1998) and bounds on the error probability were obtained. In the numerical examples these bounds were seen to converge. This paper examines analytically the behavior of the bounds for high signal-to-noise ratio (SNR) scenarios. Slowly fading channels, where the fading coefficients are essentially constant over two successive symbol intervals, and fast fading channels, where the fading coefficients can vary from one symbol interval to the next, are considered. For slow fading, the asymptotic convergence of the upper bound to the lower bound is proved. The asymptote that is reached for high values of the SNR does not depend on the interfering users' energies, establishing thereby the near-far resistance of the optimum multiuser DPSK detector. For fast fading the error floor reached for high SNRs is bounded from below and above.