Showing papers in "Bell System Technical Journal in 1966"

Bounds for certain multiprocessing anomalies

Abstract: It is known that in multiprocessing systems composed of many identical processing units operating in parallel, certain timing anomalies may occur; e.g., an increase in the number of processing units can cause an increase in the total length of time needed to process a fixed set of tasks. In this paper, precise bounds are derived for several anomalies of this type.

Synthesis of band-limited orthogonal signals for multichannel data transmission

Abstract: This paper presents a principle of orthogonal multiplexing for transmitting a number of data messages simultaneously through a linear band-limited transmission medium at a maximum data rate without interchannel and intersymbol interferences. A general method is given for synthesizing an infinite number of classes of band-limited orthogonal time functions in a limited frequency band. Stated in practical terms, the method permits the synthesis of a large class of practical transmitting filter characteristics for an arbitrarily given amplitude characteristic of the transmission medium. Rectangular-shaped ideal filters are not required. The synthesis procedure is convenient. Furthermore, the amplitude and the phase characteristics of the transmitting filters can be synthesized independently. Adaptive correlation reception can be used for data processing, since the received signals remain orthogonal no matter what the phase distortion is in the transmission medium. The system provides the same signal distance protection against channel noises as if the signals of each channel were transmitted through an independent medium and intersymbol interference in each channel were eliminated by reducing data rate.

Phase vocoder

Abstract: A vocoder technique is described in which speech signals are represented by their short-time phase and amplitude spectra. A complete transmission system utilizing this approach is simulated on a digital computer. The encoding method leads to an economy in transmission bandwidth and to a means for time compression and expansion of speech signals.

Techniques for adaptive equalization of digital communication systems

Abstract: A previous paper described a simple adjustment algorithm which could be employed to set the tap gains of a transversal filter for the equalization of data transmission systems. An automatic equalizer was shown which used this algorithm during a training period of test pulse transmission prior to actual data transmission. The present paper extends the utility of this automatic equalization system by permitting it to change settings during the data transmission period in response to changes in transmission channel characteristics. Three schemes for accomplishing this adaptive equalization without the use of test signals are described and evaluated analytically. The first such scheme uses periodic estimates of channel response based on the received data signal to adjust or update the transversal filter settings. The second system is entirely digital and employs a sequential testing procedure to make adjustments aperiodically as they are required by changing conditions. The third system uses information obtained from a forward-acting error correction system for the purposes of adaptive equalization. Of the three systems described, the second is not only theoretically superior, but is practically the simplest. Experimental results for this second system are described.

Predictive quantizing systems (differential pulse code modulation) for the transmission of television signals

Abstract: Differential pulse code modulation (DPCM) and predictive quantizing are two names for a technique used to encode analog signals into digital pulses suitable for transmission over binary channels. It is the purpose of this paper to determine what kind of performance can be expected from well-designed systems of this type when used to encode television signals. Systems using both previous sample and previous line feedback are considered. A procedure is presented for the design of nonadaptive, time invariant systems which are near optimum in the sense that the resulting signal to unweighted quantizing noise ratios (S/N) are nearly maximum. Simple formulas are derived for these S/N ratios which apply to DPCM as well as standard PCM. Standard PCM is shown to be a special case of DPCM. These formulas are verified by digital computer simulation. Any advantage of DPCM stems from removing the redundancy from the signal to be transmitted. Redundancy in a signal, however, affords a certain protection against noise introduced in the transmission medium. The penalty for removing this redundancy, through DPCM or other means, is that the transmitted signal becomes more fragile and requires a higher-quality transmission medium than would otherwise be required. This penalty is discussed in quantitative terms.

Beam-lead technology

Abstract: A process has been developed to batch-fabricate beam-lead transistors, integrated circuits, and other components, where the leads serve a structural and protective as well as electrical function. Platinum silicide ohmic contacts, titanium and platinum sputtered layers, and electroformed gold beam leads constitute the metallurgical structure of the devices described. Test transistors have survived 350°C aging for hundreds of hours in corrosive ambients, and centrifuging at 135,000 g's.

Difference-set cyclic codes

Abstract: Codes exist which are capable of correcting large numbers of random errors Such codes are rarely used in practical data transmission systems, however, because the equipment necessary to realize their capabilities — that is, to actually correct the errors — is usually prohibitively complex and expensive The problem of finding simply implemented decoding algorithms or, equivalently, codes which can be decoded simply with existing methods, is perhaps the outstanding unsolved problem in coding theory today In this paper, a new class of random-error-correcting cyclic codes is defined These codes have two very desirable features: the binary members of the class are nearly as powerful as the best-known codes in the range of interest, and they can be decoded with the simplest known decoding algorithm Unfortunately there are relatively few codes with useful parameters in this class, despite the fact that the class is infinite

Signal-to-noise and idle channel performance of differential pulse code modulation systems – particular applications to voice signals

Abstract: Analysis leading to a figure of merit for differential pulse code modulation (DPCM) systems with linear feedback networks is presented. It is shown that the figure of merit can be optimized. Simple DPCM has a 6-dB advantage in signal/quantizing noise ratio over pulse code modulation (PCM) for speech. Optimization yields at most 4 dB more. Computer simulation of the system using actual speech samples leads to data supporting the figure of merit as a useful measure of performance for DPCM systems with four digits or more. The simulation also provides data on the error spectrum as a function of quantizer loading and on the probability density of the quantizer input as a function of loading. Performance of the optimum system as a function of increasing feedback network complexity is also shown. Idle channel performance of a particular system is analyzed, indicating the presence of inband oscillations in many cases. The best quantizer bias from the point of view of idle channel performance is found. The level of idle channel noise in DPCM is shown to be approximately equivalent to that in PCM.

Topology of thin film RC circuits

Abstract: Integrated RC circuits can be made by depositing exceedingly thin metallic and dielectric films in suitable patterns on an insulating substrate. Resistors are strips of conductor; capacitors are patches on which conducting, dielectric, and conducting layers are superimposed. Since conductors can cross at capacitor patches, RC networks need not be strictly planar to be realizable in thin film. Determining which RC circuits are realizable poses new problems in topology which are remarkably simple to state but are as yet unsolved. The results reported here are fragmentary, but they do cover some cases of small order that may be of practical interest.

Ideal mos curves for silicon

Abstract: Ideal curves of MOS capacity and surface potential are computed for silicon with oxide thickness and doping as a parameter. High-frequency and low-frequency capacity curves are presented for the doping ranges between 1 × 1014 and 1 × 1017 cm−3 and SiO 2 thickness between 100 and 12000 A. Additional curves give flatband capacitance, minimum capacitance and voltage of the minimum capacitance in the same ranges.∗

Delta modulation quantizing noise analytical and computer simulation results for Gaussian and television input signals

Abstract: Quantizing noise in delta modulation systems falls into two categories, granular noise and slope overload noise. Granular noise exists because the decoded output signal can assume only a specified number of levels in the range of interest. Slope overload noise occurs when the slope of the input signal is greater than the delta modulator is capable of reproducing. When the S/N ratio is not too small, the noise power from these two sources is additive. A formula for the slope overload noise power for Gaussian input signals is derived. This is used with an earlier result for the granular noise to give over-all signal-to-quantizing-noise ratios. It is shown by computer simulation that the assumptions employed in deriving these signal-to-quantizing-noise ratios are valid and that the analytical results provide good estimates of the true values of these ratios. Computer simulation of the Picturephone∗ visual telephone, a low-bandwidth television system, illustrates that the formulas derived for Gaussian signals apply quite well to video signals. Estimates are given for the S/N ratios theoretically possible when a 4.5-mc monochrome television signal is transmitted by delta modulation. The characteristics of delta modulation quantizing noise may cause it to be subjectively less annoying than an equivalent amount of additive Gaussian noise.

A model for the random video process

Abstract: For problems concerning the transmission of video signals, it is often desirable to know the statistical distribution of power in the frequency domain for the signal process. It is convenient to have a model, involving only a few essential parameters, which will satisfactorily characterize the power spectral density of the random video signal. This paper proposes a model for the random picture and derives expressions for second-order statistical properties of the video signal obtained from a conventional scanning operation on the picture. The properties of typical picture material make valid certain approximations which lead to especially simple, closed-form expressions for power spectral density. The continuous part of the power spectral density is expressed as a product of three factors, characterizing separately the influence of point-to-point, line-to-line, and frame-to-frame correlation. For parameters representative of typical picture material there is observed an extreme concentration of power near multiples of the line scan and frame scan rates. An illustrative example of the use of the model in an optimum linear filtering problem is included.

Theory of error rates for digital fm

Abstract: A general theory is presented for evaluating the error performance of a digital FM system in the presence of additive noise. The digital system considered is a conventional one employing a voltage-controlled oscillator as the modulator and a limiter-discriminator followed by a low-pass filter as the demodulator. Because of the nonlinear nature of the demodulation process, no adequate analytical techniques have been available to provide a satisfactory treatment. Adopting the notion of “clicks” used by S. O. Rice to study threshold effects in analog FM systems, we have succeeded in evolving a theory capable of predicting performance for a wide range of applications. While our theory reinforces some previously derived results for binary and for narrow-band systems, the results obtained here are not confined to these situations. In particular, the inefficiency of the FM discriminator as a detector for a large number of orthogonal signals is quantitatively evaluated, as well as the role of the post-detection filter. Some qualitative aspects of the error-causing mechanisms discussed in the paper are general, but quantitative results are confined to additive Gaussian noise and large signal-to-noise ratios.

Mutual synchronization of geographically separated oscillators

Abstract: A control scheme for synchronizing the frequencies of geographically separated oscillators connected by communication links consists of averaging the phases received at each station from remote oscillators, comparing the result with the local phase, and applying the filtered error signal as a correction to the local oscillator frequency. The system was studied by V. E. Benes who found a sufficient condition for the stability of the system using advanced mathematical techniques. In this paper, the stability condition is derived (for a slightly more general control scheme) using only the transfer function concept of linear systems and some properties of determinants. A practical difficulty regarding the final frequency of the oscillators is discussed and a modification of the control scheme is shown to alleviate the difficulty. Also examined are the questions of sensitivity to parameter changes, the effect of jitter noise on the performance of the system, and the effect of failure of an oscillator or transmission link.

The capacity of the band-limited Gaussian channel

Abstract: Shannon's celebrated formula W ln(1 + P o /N o W) for the capacity of a time-continuous communication channel with bandwidth W cps, average signal power P o , and additive Gaussian noise with flat spectral density N o has never been justified by a coding theorem (and “converse”). Such a theorem is necessary to establish W ln(1 + P o /N o W) as the supremum of those transmission rates at which one may communicate over this channel with arbitrarily high reliability as the coding and decoding delay becomes large. In this paper, a number of physically consistent models for this time-continuous channel are proposed. For each model the capacity is established as W ln(1 + P o /N o W) by means of a coding theorem and converse.

Schottky barrier photodiodes with antireflection coating

Abstract: Schottky barrier diodes can be used for fast and efficient photodetectors if the incident light is coupled into the depletion layer of the diode and if electron-hole pairs are created by the internal photoelectric effect in the depletion layer Fast response of the diode is achieved by designing a Schottky barrier with a small RC product High quantum efficiency is obtained by coupling the light through a thin metal layer into the depletion region of the diode and by using an antireflection coating on the metal layer for matching the incident light beam Schottky barrier photodiodes have been made with thin semitransparent gold layers on n-type epitaxial silicon and with zinc sulfide as an antireflection coating A net quantum efficiency of 70 percent has been achieved at the He-Ne laser wavelength of 6328 A The pulse response of packaged diodes with 05-nanosecond wide pulses shows a symmetrical pulse shape with only small distortion due to carrier diffusion and reactance in the completed package The diode structure is suitable for detector arrays It is also useful for optical time domain reflectometry The technique of coupling light through metal layers can be extended to other optical devices which require efficient transfer of radiation into a semiconductor through conducting electrodes

A self-adaptive echo canceller

Abstract: Conventional echo suppressors combat echoes generated at hybrid junctions in long distance telephone connections by interrupting the return path according to some decision scheme based upon the relative levels of the outgoing and return signals. In this brief, a new device is described for cancelling the echo without interrupting the return path. We call this device an echo canceller to distinguish it from conventional echo suppressors. It generates a replica of the echo (which is then subtracted from the return signal) by synthesizing a linear approximation to the echo transmission path. It is self-adapting in that it automatically tracks variations in the echo path which may arise during a telephone conversation (e.g., connection or disconnection of extension phones, etc.).

Avalanche region of impatt diodes

Abstract: The avalanche region of an IMPATT (IMPact ionization Avalanche Transit Time) diode under small signal conditions is characterized by the fraction of the total alternating current that is carried by holes and electrons in their respective drift spaces and by a residual impedance. The current fractions are roughly in phase with the total current below, and nearly 180° out of phase above a resonance frequency that is proportional to the square root of direct current density. This paper extends the calculations of Gilden and Hines for the current fraction to include phase shifts in the avalanche region so that extended avalanche regions can be considered. Realistic values (α ≠ β in Si) for the ionization coefficients are used. Results of detailed numerical calculations for the current fractions as a function of frequency and direct current density are presented. For typical frequencies and current densities, the residual impedance is negligible and hole and electron current fractions are equal. The avalanche region at a given frequency and current density is then characterized by one complex number and the admittance of a diode containing the avalanche region and adjacent drift regions is easily calculated. Plots showing the admittance as a function of frequency and current density for typical structures are given. It is found that an optimal exponential growth rate of oscillations is obtained when the current density is such that the resonance frequency is about equal to one half the reciprocal transit time through the longest drift region. If the assumption is made that conditions giving the largest small-signal exponential growth rate give the best large-signal performance, an optimum Read-diode design is obtained for which the avalanche region width is a substantial fraction (≍ 1/3) of the drift region width.

B.S.T.J. briefs Multicolor holographic image reconstruction with white-light illumination

Abstract: Color images have been obtained by wavefront reconstruction from a reflection volume hologram illuminated with ordinary white light. The hologram was recorded with coherent light at two wavelengths, 6328 A and 4880 A, from helium-neon and argon-ion lasers, respectively. Fig. 1 shows the white-light reconstructed image from such a hologram; the original subject was a color transparency. The hologram was formed in Kodak 649F emulsion.

On the number of information symbols in difference-set cyclic codes

Abstract: The concept of a difference-set cyclic code has been described previously. It was shown that such a code is almost as powerful as a Bose-Chaudhuri code and considerably simpler to implement. It is the purpose of this paper to determine some of the more important properties of this code and its dual code (cf. Sec. IV). It may be pointed out that the problems we consider are equivalent to determining certain properties of incidence matrices associated with a class of balanced incomplete block designs formed from simple difference sets.

B.S.T.J. briefs Hologram transmission via television

Abstract: Holography, or wavefront reconstruction photography, was first demonstrated by Gabor1,2,3 over fifteen years ago, and it has been the subject of increased investigation over the last five years since the advent of lasers. Possible applications of holography suggest themselves in the fields of three-dimensional and multicolor television.4 Furthermore, the statistics and redundancy of the hologram of an image may be deliberately made quite different from those of the original image. This has obvious possibilities in encoding schemes for television transmission in general. In this communication we report a first experimental step in this direction, namely the successful transmission via television of a Fresnel type of hologram in which the original object was a transparency.

Sequential decoding — the computation problem

Abstract: Sequential decoding is a technique for encoding and decoding at moderate cost with a decoding reliability which approximates that of the optimum, and expensive, maximum-likelihood decoder. The several known sequential decoding algorithms enjoy a cost advantage over the maximum-likelihood decoder because they allow the level of the channel noise to regulate the level of the decoding computation. Since the average level of the required decoding computation for sequential decoders is small for source rates below a rate R comp , such a decoder can be realized for these rates with a relatively small logic unit and a buffer. The logic unit is normally designed to handle computation rates which are less than two or three times the average computation rate; the buffer serves to store data during those noisy periods when the required computation rate exceeds the computation rate of the logic unit. If the periods of high computation, which are caused by noise, are too frequent or too long, the buffer, which is necessarily finite in capacity, will fill and overflow. Since data are lost during an overflow, continuity in the decoding process cannot be maintained. The decoder, then, cannot continue to decode without error. For this reason, buffer overflow is an important event. In addition, since errors in the absence of overflow are much less frequent than are overflows themselves, the overflow event is of primary concern in the design of a sequential decoder. This paper presents some recent analytical results concerning the probability of a buffer overflow. In particular, it is shown that this probability is relatively insensitive to both the buffer capacity and the maximum speed of the logic unit for moderate capacities and speeds. By contrast, it is shown that the overflow probability decreases rapidly with a decrease in the source

Optics of general guiding media

Abstract: Weakly focusing transparent media provide possible means for guided transmission of coherent light beams with relatively small loss. The scalar wave equation for the eigenmodes of propagation in such a medium is formally identical with Schroedinger's wave equation. Hence, the methods used in the solution of quantum-mechanical problems, such as the Wentzel-Kramers-Brillouin (WKB) approximation, are immediately applicable to this problem. Solutions for the eigenmodes and eigenvalues in the case of focusing in one dimension are given, and the Poschl-Teller medium, whose index varies as n = n 0 [1 − (α/2) tan2 ηx] is discussed in some detail. In addition, the relationship between the wave solutions and geometrical (ray) optics is examined.

Bounds on communication with polyphase coding

Abstract: The theoretical capabilities of a “polyphase” coding-modulation scheme with additive white Gaussian noise are studied. The channel capacity of this system is found and the error exponent estimated. Bounds are also found on R o (ρ max ), the maximum (asymptotic) rate for which polyphase codes can be found with maximum correlation between code words ρ max .

Geometrical representation of Gaussian beam propagation

Abstract: A single equation governing the various properties encountered in propagation of Gaussian beams is discussed. These characteristics may be graphically presented on an impedance chart such as a Smith chart or in the form of normalized curves. The geometrical representations highlight the relation between Gaussian mode theory and geometrical optics formulas. The power coupling coefficient between two Gaussian modes suggests a complex mismatch coefficient whose geometrical representation is essentially the same as that of a complex reflection coefficient in transmission-line theory. Application of the Smith chart in determining a complex mismatch coefficient is illustrated by graphical solution of a beam-matching problem.

Application of automatic transversal filters to the problem of echo suppression

Abstract: Long-haul voice communication has long been subject to the problem of returned echo. The advent of synchronous satellite communication introduces increased delay as a degrading factor in the overall quality of two-way conversation. This compounds the problem in that the echo

B.S.T.J. briefs Display of holograms in white light

Abstract: This paper describes a new method for displaying holograms in white light. The method gives reasonably good reconstructions although certain image defects are inherent in the method. It differs from previously reported methods of white light reconstruction1,2 in that the whole spectrum is used for reconstruction and therefore black and white reconstructions can be obtained. The method does not depend on the volume properties of the photographic emulsion.

Theory of cascaded structures: Lossless transmission lines

Abstract: Cascaded structures play a major role in many signal processing and signal propagating systems. The universality of such structures is particularly evident when the signals are of a wave nature, i.e., the components of the structure are representable by transmission lines rather than lumped elements. Transmission media with discontinuities are examples of such structures. Other examples include integrated, microwave, and optical circuits. Theory of distributed structures has, so far, been successfully developed only for structures whose components are lossless (or RC) transmission lines of equal electrical lengths. It is the purpose of this paper to present a theory of cascaded structures when the component elements are lossless transmission lines of arbitrary electrical lengths. Extensions of the theory developed here to other structures will be discussed in a subsequent paper.

Programming and control problems arising from optimal routing in telephone networks

Abstract: In many circumstances a telephone call can be completed through a connecting network in several ways. Hence, there naturally arise problems of optimal routing, that is, of making the choices of routes so as to achieve extrema of one or more measures of system performance, such as the loss (probability of blocking) or the carried load. As is customary in traffic theory, a Markov process is used to describe network operation with complete information. The controlled system is described by linear differential equations with the control functions (expressing the routing method being used) among the coefficients. Restricting attention to asymptotic behavior leads to a problem of maximizing a bilinear form subject to a linear equality constraint whose matrix is itself constrained to lie in a given convex set. An alternative approach first shows that minimizing the loss, and maximizing the fraction of events that are successful attempts to place a call, are equivalent. This fact permits a dynamic programming formulation, which, in turn, leads to a very large linear programming problem. Two small examples are treated numerically by this method. It is particularly important to try to verbalize, and then mechanize, the optimal routing strategies. In this endeavor, the linear programming formulation is of limited usefulness. Therefore, in the latter half of the work we

The determination of frequency in systems of mutually synchronized oscillators

Abstract: The synchronization of large systems of geographically separated oscillators is of considerable practical interest for pulse code modulation (PCM) switching. This study examines the factors that determine the frequency at which such a system operates, considering both the procedure by which it is set up and the topology of system interconnections. A necessary and sufficient connectivity condition is established.