This paper presents serially concatenated trellis coded modulations (SCTCMs) that perform consistently close to the available mutual information for periodic erasure channel (PEC), periodic fading channel (PFC) and the 2 times 2 compound matrix channel. We use both the maximum-likelihood decoding criteria and iterative decoding criteria to design universal SCTCMs for the PEC and the PFC. For the space-time channel, by demultiplexing the symbols across the antennas, the proposed universal SCTCMs for the period-2 PFC deliver consistent performance over the eigenvalue skew of the matrix channel. Within the family of channels having the same eigenvalue skew, a time-varying linear transformation (TVLT) is used to mitigate the performance variation over different eigenvectors. The proposed space-time SCTCMs of 1.0, 2.0 and 3.0 bits per transmission require excess mutual information in the ranges 0.11-0.15, 0.23- 0.26 and 0.35-0.53 bits per antenna, respectively. Because of their consistent performance over all channels, the proposed codes will have good frame-error-rate (FER) performance over any quasi-static fading distribution. In particular, the codes provide competitive FER performance in quasi-static Rayleigh fading.

In broadcast scenarios or in the absence of accurate channel probability distribution information, code design for consistent channel-by-channel performance, rather than average performance over a channel distribution, may be desirable. Root and Varaiya's compound channel theorem for linear Gaussian channels promises the existence of universal codes that operate reliably whenever the channel mutual information (MI) is above the transmitted rate. This paper presents 2-D trellis codes that provide such universal performance over the compound linear vector Gaussian channel when demultiplexed over two, three, and four transmit antennas. The presented trellis codes are found by an exhaustive search that guarantees consistent performance on every matrix channel that supports the information transmission rate with an MI gap that is similar to the capacity gap of a well-designed additive white Gaussian noise (AWGN)-specific code on the AWGN channel. As a result of their channel-by-channel consistency, the universal trellis codes presented here also deliver comparable, or in some cases, superior frame-error rate and bit-error rate performance under quasi-static Rayleigh fading to trellis codes of similar complexity that are designed specifically for the quasi-static Rayleigh fading scenario.

In broadcast scenarios or in the absence of accurate channel probability distribution information, code design for consistent channel-by-channel performance, rather than average performance over a channel distribution, may be desirable. Root and Varaiya's compound channel theorem for linear Gaussian channels promises the existence of universal codes that operate reliably whenever the channel mutual information (MI) is above the transmitted rate. This paper presents two-dimensional trellis codes that provide such universal performance over the compound linear vector Gaussian channel when demultiplexed over two, three, and four transmit antennas. The presented trellis codes, found by exhaustive search, guarantee consistent performance on every matrix channel that supports the information transmission rate with an MI gap that is similar to the capacity gap of a well-designed additive white Gaussian noise (AWGN)-specific code on the AWGN channel. As a result of their channel-by-channel consistency, the universal trellis codes presented here also deliver comparable, or, in some cases, superior, frame-error rate and bit-error rate performance under quasi-static Rayleigh fading, as compared with trellis codes of similar complexity that are designed specifically for the quasi-static Rayleigh-fading scenario

In the absence of accurate channel probability distribution information or in broadcast scenarios, code design for consistent channel-by-channel performance, rather than average performance, may be desirable. Root and Varaiya's compound channel theorem promises the existence of universal codes that operate with a consistent proximity to channel mutual information on any instance of the compound linear vector Gaussian channel that is similar to the capacity gap of an AWGN-specific code with similar complexity on the AWGN channel. This study presents single-dimensional trellis codes such that when multiplexed over two, three and four transmit antennas, provide universal performance over the compound linear vector Gaussian channel. As a result of their channel-by-channel consistency, the universal trellis codes presented here deliver comparable or in some cases superior frame-error-rate and bit-error-rate performance under quasistatic Rayleigh fading to trellis codes of similar complexity that are designed specifically for the quasistatic Rayleigh fading scenario.

This paper extends the near-capacity performance of serially-concatenated convolutional codes under AWGN to periodically time-varying channels. In particular, we propose serially-concatenated trellis-coded modulations (SCTCMs) that perform consistently close to channel capacity under periodic erasures as well as under AWGN, without sacrificing AWGN channel performance. The proposed 0.5 bits/symbol and 1.5 bits/symbol SCTCM schemes are robust with respect to periodic time-variations with a period of two-symbols.

This article gives practical examples of space-time trellis codes performing as predicted by Root and Varaiya's (1968) compound channel theorem. Specifically, 32-state and 64-state 2/spl times/2 space-time trellis codes are presented that provide a bit-error rate (BER) of 10/sup -5/ on all 2/spl times/2 matrix channels with an excess mutual information (MI) within 8% of the excess MI required by standard trellis codes of the same complexity operating only in additive white Gaussian noise (AWGN). Not surprisingly, the universal space-time trellis-coded modulations (ST-TCMs) provide average bit- and frame-error rates in quasi-static Rayleigh fading (QRF) that are comparable to those achieved by ST-TCMs designed specifically for QRF as well as standard TCMs followed by the Alamouti (1998) space-time block code. However, all of these other schemes require more excess MI in the worst case, and some have a significantly wider variation in the required excess MI. The article also compares the universal and quasi-static Rayleigh fading design approaches analytically and bounds the worst case distance of a trellis code on a 2/spl times/2 channel using the distances of the code on singular and unitary channels. This bound is extended to the more general n/sub T//spl times/n/sub R/ scenario.

The compound channel theorem for linear Gaussian vector channels indicates that a single code can provide reliable communication on all channels that induce a minimum amount of mutual information. This paper discusses the search for space-time trellis codes that deliver such consistently good performance on every matrix channel that has a proximity to channel capacity similar to standard trellis codes in AWGN. By characterizing the minimum distance of an error event transformed by the channel as a function of the eigenvalues of the channel matrix, we search for codes having the largest such worst-case distance for systems with two transmit antennas. Proposed trellis codes found by exhaustive search guarantee coded performance on every matrix channel without sacrificing error performance under quasistatic Rayleigh fading.

This paper presents a worst-case pairwise codeword error probability analysis, under arbitrary block fading, for wireless communication systems with multiple transmit and receive antennas. Our results generalize the Rayleigh-fading space-time code design criteria (full rank and maximum determinant) to arbitrary fading.

Previous work in sequential multihypothesis testing has considered the goal of minimizing the expected number of observations required to choose a hypothesis with a desired level of accuracy. Motivated by reduced-complexity decoding applications, we consider sequential multihypothesis testing techniques that remove individual hypotheses from consideration as they become unlikely in order to minimize the expected aggregate number of hypotheses tested. In the limit of small decision error probabilities, the optimal sequential test that rejects a single hypothesis is characterized. A full minimum complexity sequential multihypothesis testing scheme which assumes the same error probability at each drop of a hypothesis then follows in a straightforward manner. Numerical results are presented that demonstrate the complexity savings via this approach for two simple examples.

The Shannon capacity of a fading channel under an average-power constraint with channel side information at the transmitter and receiver is only negligibly larger than the capacity of the same channel when constant-power transmission is employed. However, power adaptation has been shown to be quite useful in practical systems, where it has been conjectured that it allows for compensation of the effect of rate quantization. Here, an average bit-error probability constraint is employed instead of the conventional instantaneous bit-error probability constraint. When the set of rates available to the transmitter is unrestricted in practical systems, necessary conditions for jointly optimal power and rate allocation are derived and used to demonstrate that power adaptation is of limited utility. However, when the rates available to the transmitter are restricted to the nonnegative integers for the example of uncoded quadrature amplitude modulation over frequency-nonselective Rayleigh fading channels, a 0.5-0.75 dB loss in power efficiency is incurred when employing only a single power level for each constellation, and a 0.5-bits/symbol loss in rate is incurred when constant power transmission is employed.

We develop a technique to perform efficient and accurate matching to detect the occurence(s) of a template in a scene. The template may describe an object or a textured surface. Representative binary features such as edge points for objects and a set of intensity extrema for textured surfaces are extracted from both test and template images. The search is based on a three-stage "coarse-fine-final" matching. First, we match with the binarized template skipping positions in horizontal and vertical directions and find a number of highest ranking matching positions. Second, we search in neighborhoods centered at the stated positions and find improved matching positions. Finally, some error measure between the original, i.e. gray level or color, test and template images is computed at the fine positions as the final matching. It is demonstrated that the proposed scheme achieves significant reduction in computational complexity with virtually no loss of detection performance.