Presents the welcome message from the conference proceedings.

We describe a system, a wireless overlay on existing hybrid fiber/coax digital packet-based networks, for high-speed, short-range wireless access to video programming services, to the Internet, and to the telephone, cellular mobile, and other communication networks. Wireless access may be provided to both fixed (residential) users and to passersby with portable devices. Smart transponder nodes and antennas are densely distributed along the access cable line for wireless communication with user nodes. The antennas along a cable access line form an adaptive "superantenna" that is excited, for downstream transmission, by short sequences of vector signals that cancel radio interference. These excitations can be derived from a stable adaptive process. Upstream transmission from a subscriber node is designed so that the vector received by the superantenna can be used to separate this signal from that of other users' transmitters. Cooperation among end users is not necessary but is possible. Antenna excitation signals are controlled from the system headend, based on processing that uses measured channel parameters between the sets of access line and user antennas. A program controller in the headend, responding to end user requests, manages delivery of selected program streams and other customer services through generation of superimposed superantenna excitation signals that target multiple end users. This model offers opportunities for flexible capacity allocations and a wide range of programming and customized private.

Summary form only given, as follows. Our three publications - Communications Magazine, IEEE Transactions on Communications, and the Journal on Selected Areas in Communications - cover an area of technology that is surely one of the most exciting and exuberant today. Electronic communications makes headlines in the daily press, creates whole new industries, and has an ever-increasing effect on everyone's life. The technical details captured in our pages are, or at least should be, closely tied to the strong currents of technological change which are sweeping through society. In order to capture thjs spirit of newness and excitement, our publications need to serve our readers by staking out claims in the new areas and publishing more in those areas. Personal computer communications, the role of electronic countermeasures in recent international events, the revolution in device fabrication for communications applications, new optical fiber communications systems, and the new products and services brought on by communication deregulation are all "hot" topics which should be given more emphasis, and will be. We would welcome your additions to this list. On the other hand, there are constraints on this process. The Communications Society spends more money on its publications than on anything else. Thus, it is unavoidable that we publish less on subjects which are old and overworked and do not represent the future. Our finite page limits compel us to be selective and to emphasize papers which are both of high quality and in the ground-breaking subject areas. The three of us are asking our editors and reviewers to impose a stricter standard of significance on papers so as to emphasize those that make a sizeable increment in the state of the art or wnich favor new topics over old ones. Higher rejection ratios may be expected in some areas. We want to hear from you how this is working, because feedback from readers is the most desirable and reliable guide for our editorial policies. Please do not expect any large immediate changes in the content of the publications. Yet, we hope that within six months you will begin to find the articles ·more interesting, relevant, and original, and above all that you are actually reading more of them. This will be the real test of our new policies.

A digitally-implemented fractionally-spaced equalizer will generally exhibit long-term instability when the conventional tap-adjustment algorithm is used. This occurs because, in contrast to the synchronous equalizer, a fractionally-spaced equalizer will generally have many sets of tap values which result in nearly equal values of mean-squared error. Some of these tap settings are large enough to cause register overflows and consequent performance deterioration. In this paper we report how a simple modification in the tap-adjustment algorithm provides a solution to the above problem.

Adaptive mean-square-tapped-delay-line echo cancellers for voice applications are conventionally designed to stop adjustment during periods of "double-talking", i.e., when a large informationbearing signal is present along with the echo signal to be cancelled. Continuous adaption is, however, desirable in full-duplex, two-wire data transmission where the periods of double-talking are so long that the echo channel may vary. We presume that the tap weights of an echo canceller have converged during a training period free of double talking, and address the problem of subsequent echo-canceller tap adjustment via the estimated-gradient algorithm in the presence of double talking. In the estimated-gradient algorithm the tap increment should be proportional to the product of the residual echo and the tap voltage. However, when double talking occurs the residual echo can only be estimated. For an idealized double-talking model, it is demonstrated, from infinite-precision considerations, that use of the memoryless maximum-likelihood estimate of the residual echo is nearly equivalent to abrupt reduction of the step size of the adjustment algorithm when double-talking begins, and could provide an automatic mechanism for recognizing double-talking. Unfortunately, the response of a digitally implemented canceller to a sharply reduced step size can be a deterioration in performance. In fact, the use of an exceedingly small step size during periods of doubletalking may lead to a cancellation error considerably larger than that predicted by coefficient precision. It is demonstrated how averaging the estimated gradient can significantly decrease the mean-squared tap error during periods of double talking. To a first approximation, the tap-weight error can be reduced by a factor proportional to the averaging interval, with an equivalent decrease in tracking capability.

A digitally-implemented echo canceller operating at a rate greater than twice the highest passband frequency is proposed for fullduplex data transmission on a two-wire circuit. Located at each end of the communication circuit, the cancellers operate independently of the local receivers and do not require synchronization of the two data stations. Economies in storage and A/D conversion, in comparison with voice-type cancellers described in the literature, are achieved by accepting data symbols as reference input instead of samples of the transmitted line signal. Convergence of transversal filter tap weights is demonstrated under a mean-squared error criterion, and use of the real passband error rather than the complex analytic error is found to lead to the same residual error at the expense of a doubled convergence time. An operational protocol and adaptation algorithm are proposed which make possible both rapid start-up and slower adaptation during double talking. Provision is made for limiting the number and location of active taps on the transversal filter to those actually necessary for replicating the echo channel, resulting in two transversal filter sections of moderate length separated by a bulk delay. Results from a computer simulation of the proposed canceller are offered to demonstrate that convergence of mean-squared tap-weight error follows the predicted exponential characteristic, that the length of the bulk delay can be determined from a single channel sounding under typical channel noise conditions, and that the use of an averaged-gradient algorithm will allow the canceller to adapt, although slowly, to a change in the echo channel during full-duplex operation.

The degree of complexity of a digital signal processor is closely related to the precision with which samples of an incoming analog waveform are represented. There is considerable interest in determining how coarse this representation can be without seriously degrading performance from that of an ideal processor of unquantized samples. This question is examined for a receiver of noisy, linearly distorted pulse amplitude modulation (PAM) signals. An optimum [maximum likelihood (ML)] detector, analogous to the Viterbi detector for unquantized samples, is derived for the case of a quantized sample sequence. Performance is evaluated under the assumption of high signal-to-noise ratio (SNR), and the resultant error probability is a good approximation for coarse quantization, and an upper bound for any degree of quantization. For a specified error probability, the degree of quantization suggested by this approach is conservative. Since receiver complexity is closely associated with the length of the digital representation of an input sample, an upper bound on receiver complexity is also suggested. Numerical evaluation of the error probability is quite tedious for an arbitrary channel; however, system performance may be readily evaluated for partial-response (PR) signaling. For the PR channels

It is shown that under certain moderate conditions, a large class of passband data signals, including all two-dimensional signal sets and many multilevel frequency-shift keying (FSK) signals, can be constructed from a finite set of stored waveforms. A description is given of a transversal filter with time-varying tap weights which can generate this class of signals. The configuration is amenable to digital implementation and to use in program-controlled systems.

A considerable amount of literature exists on the problem of selecting an efficient set ofNdigital signals with in-phase and quadrature components for use in a suppressed carrier data transmission system. However, the signal constellation which minimizes the probability of error in Gaussian noise, under an average power constraint, has not been determined when the number of signals is greater than two. In this paper an asymptotic (large signal-to-noise ratio) expression, of the minimum distance type, is derived for the error rate. Using this expression, a gradient-search procedure, which is initiated from several randomly chosenN-point arrays, converges in each case to a locally optimum constellation. The algorithm incorporates a radial contraction technique to meet the average signal power constraint. The best solutions are described for several values ofNand compared with well-known signal formats. As an example, the best locally optimum 16-point constellation shows an advantage of about 0.5 dB in signal-signal-to-noise ratio over quadrature amplitude modulation. The locally optimum constellations are the vertices of a trellis of (almost) equilateral triangles. AsN \rightarrow \infty, it is rigorously proved in the Appendix that the optimum constellations tend toward an equilateral structure, and become uniformly distributed in a circle.

In-band generation of synchronous linear data signals eliminates the need for preliminary generation of a baseband signal and sharp cutoff filtering. However, the ratio of carrier frequency to pulsing rate must, in general, be an integer if the signal is to be produced by successive pulsing of a single fixed filter. As a consequence, a singlesideband signal of bandwidthWand transmission rate2Wmust occupy the frequency band [2NW,(2N +1)W],Nan integer, when it is generated from a single fixed filter. A partial-response signal satisfying this condition is described in this paper. A further analysis demonstrates that a finite number of fixed filters can be used to produce a pulse train in any band where the ratio of carrier frequency to pulsing rate is a rational number. Since this relaxed condition allows any transmission band to be approximated to an arbitrary degree of accuracy, frequency translation can be avoided. A single transversal filter with periodically varied tap weights is shown to be a practical realization.

Simultaneous two-way (full-duplex) data transmission in the same bandwidth normally requires separate channels, and in the telephone network is realized by using a four-wire circuit. This paper reports on investigations directed toward full-duplex operation on a two-wire circuit using a hybrid coupler and a passband automatic transversal filter at each end of the circuit to separate the two signals. Full-duplex binary transmission at a rate of 2000 Bd in each direction has been experimentally demonstrated on a limited number of twowire circuits. Photographs of eye patterns are provided to illustrate the substantial improvement over operation with hybrids alone. Limitations on performance due to unstable distant echoes are described.

An introduction is provided to the classical theory of asymptotic distributions of extreme values and to its application to estimation of tail probabilities. It is shown how the theory is sometimes misinterpreted to justify heuristic estimation procedures that introduce systematic error. A new generalized asymptotically convergent sequence of extreme-value distributions, differing from the classical sequence in higher order terms, which eventually disappear, is derived for the "exponential'' class of distributions. It is shown that in some cases of practical interest, the generalized sequence converges much faster to the asymptote than does the widely used classical sequence, and that tail-probability estimates of a given accuracy can be formulated from a much smaller number of data.