Search

The current global telecommunications plant, an estimated 50 billion dollar investment, is built around the 4-kc channel and frequency division multiplex (FDM). The characteristics of this 4-kc channel are uniquely suited to telephone communications at continental distances (mean of 2000 miles with a dynamic range varying from 200 to 6000 miles) and all but ignore predominately digital communications. The characteristics of the present plant also ignore the possibility of a high percentage of intercontinental connections (mean of 6000 miles and a dynamic range varying from 1500 to 15,000 miles). Intercontinental communication is not limited by feasibility, but by appropriate planning and generally poor signal quality, i.e., low signal-to-noise ratios for speech and high error rates for digital communication. This paper considers the characteristics of digital communications resulting from increases in intercontinental traffic, and indicates the effect of impact of these characteristics on future telecommunication networks.

The purpose of this theoretical investigation is to study the effects of fading rate on the performance of binary digital communications systems. The performance of three systems is evaluated from the aspect of probability of error per information bit when the desired and undesired signals are subject to statistically independent Rayleigh fading. Diversity reception is not included in the system comparison. The results, which apply only for a nonselective fading environment, of this investigation are the following: The effect of fading rate on communication system performance was evaluated. The information bit error probability function was explicitly related to the signal envelope and phase fluctuations. A Rayleigh-type fading with arbitrary fading rate was assumed for the signal in order to get an understanding of the system performance in a fast fading environment. Three basic types of communication systems were investigated. The first system, called the quadrature detection system, employs large-bandwidth signal duration product (WT) with quadrature detection. The next two systems employ a WT product approximately equal to one. These last two systems were the noncoherent frequency-shift keyed (FSK) and the differentially coherent phaseshift keyed (DPSK), also known as the phase comparison-type system. (Section I.) The comparison of three systems illustrating the effect of Rayleigh signal fading on the mean error probability (\bar{Pe}) showed that the phase comparison system was most critical to fast fading while the noncoherent FSK was somewhat better than the quadrature detection system. The figures show that, for a fading signal bandwidth information bit duration (BT) product of one half, the phase comparison system rapidly degraded to a \bar{P}e of one half. The other two systems showed a slower rate of degradation as the BT product increased. The quadrature detection system and the FSK receiver with incoherent detection are not identical but are equivalent for slow fading. The performance of these two systems as a function of BT is definitely different. Also, the methods of improving the performance of both systems in a fast fading environment are completely different. The graphs of error probability covering ranges of 102are of practical interest because they illustrate the comparison of the different systems as a function of fading rate and are necessary for an experimental evaluation of the systems. The exact expressions for error probability are given for evaluating any of the three system performances for ranges not covered in the graphs. (Section II.) In the error probability analysis, the basic function which is considered is the conditional probability of error per information bit averaged over the additive noise for a given desired and undesired signal envelope and phase fluctuations. This conditional probability of error will vary from one information bit to the next and, therefore, it must be considered to be a random process. Since this conditional error probability is a random process, it is important to determine its distribution function. It is shown, for the quadrature detection system, that there is a rapid degradation of the quartiles as the BT product increases, whereas the spread of the distribution decreases. An attempt was made to do a similar analysis on the other systems but the analytical problems were more difficult and time did not permit a solution. (Section III.) Received December 19, 1962. RCA Defense Electronic Products Division, Moorestown, N.J. An increase in transmitted signal power always results in a reduction in errors in the quadrature detection and FSK (filters sufficiently separated so there is no intersymbol interference) systems,but in the presence of fading there is a limit on the degree of error reduction which can be attained in the phase comparison system. In the presence of fading these theoretical results show there exists some signal level above which the error rate for the FSK system is less than that for the phase comparison-type system. These conclusions have been verified by field tests. (Section IV.) The performance of the phase comparison systems depends upon the phase stability of the propagation medium. Since these phase stabilities have not been investigated in any great detail, the complete capabilities of the phase comparison systems cannot be made for the various media. (Section V.) The shape of the spectral density of signal fading can have a significant effect on error probability performance. The single-tuned fading spectrum gave the worst error probability performance, the Gaussian spectrum was in the middle and the ideal band-pass spectrum had the least severe effect. (Section VI.) Energy detection of noncoherent FSK signals shows a very significant improvement in system performance over envelope detection in a fast fading environment. In addition, energy detection could result in better performance in a fast fading environment compared to a slow fading environment. This is completely reversed from the envelope detection system. (Section VII.) In quadrature detection using orthogonal coded-type waveforms, it is necessary to have maximum correlation over the entire waveform for minimum error probability. Fast fading will destroy this correlation and thereby increase the error probability. Improved performance may be obtained by subdividing the information bit so that the correlation of the subdivided waveform is high. The correlator outputs of this subdivided information bit are squared and added. The resulting error probability will show a significant improvement in performance in a fast fading environment compared to the nonsubdivided information bit technique. (Section VIII.)

Revolutionary developments taking place in the field of light generation show promise of providing a means for transmitting digital information over vast distances in space at extremely high rates. These developments stem from the generation of coherent light by devices called LASERS (Light Amplification by Stimulated Emission of Radiation). This paper gives a brief description of LASER operation and discusses the applicability of the device to certain aerospace vehicular digital communications requirements. An earth-moon link is an alyzed from the standpoint of beamwidth, power, and aiming requirements. It is shown that a system utilizing a coherent optical transmitter of less than 1 w and a conventional photodetector would be capable of transmitting digital information over this link at megacycle rates. The bandwidth limitation here is imposed by lack of a suitable modulator rather than by any theoretical bound, and the power level is dictated by earth background noise and a usable transmitter and receiver beam angle. Improvements which can be made on this rather simple system by increasing bandwidth and improving detection efficiency, tracking accuracy, and LASER techniques are pointed out. Curves showing basic limitations and interdependence of system parameters are plotted.

A concept of a totally digital communications network is described in which all signals are converted into digital form and remain so as they are multiplexed, switched, and transmitted. To permit time-division techniques for such network processing, and thus realize significant economic advantages, the digital signals must be either generated in synchronism or brought into synchronism. While the switching portion of this network is not yet in service, some of the transmission network already exists and the evolution toward a digital transmission hierarchy is well underway. The hierarchy consists of 1) terminals which convert analog Signals into digital form suitable for transmission, 2) transmission facilities which are available at various capacities, and 3) multiplexers which can derive several lower capacity digital facilities from a single high-capacity system. Such a network, when complete, can also disseminate time and frequency for other uses with the same accuracy as required by the network. The network requires that the relative phase difference between any two signals must be bounded, which means exact matching of the long-term average frequency throughout the network. Signal-processing time and variations in propagation velocity in various media-cable, radio, waveguide-control the short-term accuracy. This network can thus provide time-and-frequency information proportional to the observation time. Available techniques that can achieve such accuracy for the network are discussed.

The following major interrelated aspects in digital communications are described: source encoding, channel encoding and transmission, and receiver methods. Methods for digitizing speech and video are summarized. Digitized voice can be transmitted at 10 kb/s, while phone video requires 1.544 Mb/s. The major aspects of both forward-error control (FEC) and automatic request for repeats (ARQ) are considered. Power-limited forward channels are distinguished from bandwidth-limited ones. Signal distance concepts and maximum-likelihood detection principles provide the framework for describing both block-coded and convolutional-coded signaling. Convolutional coding with Viterbi decoding appears to achieve the most cost-effective forward-error control. The improved FEC techniques benefit the transmission of real-time digitized voice and video and any communication over severely power-llmited links. ARQ methods continue to be effective and practical for most cases where the delay associated with retransmission can be tolerated.

A channel identification problem arises in high speed digital communication systems that transmit information over time-dispersive channels such as telephone channels and radio channels. This paper deals with algorithms for performing the channel identification rapidly in conjunction with the signal detection to recover the information. The stability and convergence properties of the algorithms are demonstrated.

In comparing different techniques for multiplexing N binary data signals into a single channel, time-division multiplexing (TDM) is known to have a theoretic efficiency of 100 percent (neglecting sync power) and thus seems to outperform frequencydivision multiplexing systems (FDM). By considering more general FDM systems, we will show that both TDM and FDM are equivalent and have an efficiency of 100 percent. The difference between the systems is in the multiplexing and demultiplexing subsystems, but not in the performance or in the generated waveforms.

An unusual signal with dual properties has been evolved such that the waveform has one of three possible frequencies in each digit time slot as well as four phases at the transition points between the digits. Such characteristics permit the recovery of the original data input either by noncoherent FM or by differentially coherent phase detection, depending upon the characteristics of the transmission medium. At the same time, the signal affords two-to-one bandwidth compression compared to binary systems. Expressions for the constraints imposed on the system and for power spectral distribution are developed. Finally, a simple and straightforward embodiment of the concept as well as certain empirical results are presented in detail.

An upper bound is obtained on the probability that a specified pattern of errors, caused by a combination of intersymbol interference and additive Gaussian noise, occurs in an amplitudemodulated digital data transmission system. The bound, which is a generalization of Saltzberg's upper bound for the probability of a single error, behaves as an exponential function of the signal-to-noise ratio and the overall channel impulse response. The exponent of the bound for a pair of errors is calculated for two "typical" channels and in each case is fairly close to twice the corresponding Saltzberg exponent for a single error.

This tutorial paper presents an introduction to the problems of communications between aerospace vehicles and stations on the ground. The possible types of aerospace-to-ground communication links are described and classified according to the mode of signal propagation and according to obtainable range. The suitability of carrier frequencies throughout the electromagnetic radiation spectrum from very-low radio frequencies to light frequencies is discussed for the various types of links. Considerations of the information needed for the selection of modulation and demodulation techniques are presented, and factors which influence the practical value of digital coding are introduced. A discussion of the mutual influence of various parameters on the transmitter power requirement concludes the general survey.

A set of equations is derived which permits the calculation of the figure of merit for a communications channel in which a binary error correcting code is used in conjunction with correlation decoding at the receiver. The figure of merit is expressed in terms of the required probability of error and the parameters of the code. These equations are then used to determine the figure of merit for a number of systems using error correcting codes. In particular, a comparison is made between those systems which use orthogonal codes, and a class of systems using codes for which the ratio of bandwidth to information rate is equal to two.