996 PROCEEDINGS OF THE IEEE, VOL. 63, NO. 7, JULY 1975 Applications of Electrical Noise MADHUSUDAN GWA, MEMBER, IEEE Abstmcr-The omnipresent noise in electronic circuits and devices is some use for noise, but the variety of applications is often not genenlly ddered undeshbIe. This paper descdbea meof the appreciated. Furthermore, the words “applications of noise” applicrtions to which it has been put. Short descriptions of a wide are not synonymous with “applications of noise sources.” A variety of applications are given together with references for further details. The applications fall in four categories: applications in which much broader point of view is taken here to include any use to noise is used as a broad-tmnd random signal; measurements in which the which the study or measurementof noise can be put. rmdan noise is used as a teat si8p.1; measurements in which noise is The purpose of this article is threefold. First, it is intended used 18 a probe into microscopic phenomem; and the appkatior~S to generate interest in the study of noise for reasons other where noise is a theoretical tool. Many examples conceptwl or of than a performance limitation. Second, it is aimed toward applications in each of these categodes are given. Some of the rpplio as tions induded are only of historical interest now, and a few are, as yet, motivating the readers to apply random-noise techniques in dYPropoeals. their own work by providing, in summary form, the case histories of past successful applications of noise. Finally, it is I. INTRODUCTION hoped that the article would induce some fresh thinking on how noise may be employedfor other new and useful HE TITLE of this article might appear to be surprising, purposes. if notself-contradictory. For a longtime, electrical engineers have studied electrical noise (or fluctuation II. TYPESOF APPLICATIONSOF NOISE phenomena) in circuits and devices much as physicians study a For ease of study and systematization, an attempt is made in disease: not out of any affection for the subject but with a this section to classify the types of situations in which noise desire to eliminate it. In fact, noise is sometimes defined as can be made to serve a useful purpose. One possible method “any undesirable signal.” A typical chapter entitled “Noise” of listing these situations is according to the area of applica- in an electrical engineering text would usually begin by giving tion. Thus noise finds applications in biomedical engineering, such reasons for studying noise as these: “Noise contaminates circuittheory, communication systems, computers, electro- the signal, setting a limit on the minimum signal strength re- acoustics, geosciences, instrumentation, physical electronics, quired for proper communication; therefore, the reliability of reliability engineering, and other fields. A second method is to communication could be improved, power requirement could classify the applications by how noise useful rather than be reduced, or larger distances could be reached by reducing is where. This second approach will be followed here because it noise;” or, “Noise limits the accuracy of measurements, and is more fundamental and illuminating. should be minimized to improve precision.” Such statements, There are several possible characteristics of electrical noise although true, do not fully reveal the uses to which a study of which make it useful in the applications discussed here. Thus noise could be put. In this article, we propose to treat noise as there are applications based upon the fact that a noise signal a tool rather than nuisance.a can be broad-band, may arrive from a known or desired direc- To the academically oriented; the existenceof noise is reason tion in space, can have a very small amplitude, or may be un- enough to study it; it just might turn out to be useful. Most correlatedbetween nonoverlapping frequency bands. Very work on noise, however, has been motivated by more immedi- often, however, a single application is based on the existence ate applications, reduction of noise being only one of them, of several of these characteristics,making it difficult to classify although a major one. This article is devoted to some of the the applications according to these characteristics unambigu- other applications. Not very infrequently, the study of noise, ously. The noise applications are, therefore, somewhat arbi- or the use of noise as a tool in thestudy of something else, has trarilyclustered here intothe following four categories, led to significant advancements, an outstanding example being depending upon howthe noise is employed. the work of GUM [ 1 1. He was measuring noise in semicon- ductor materials, while studying the properties of hot elec- A. As a Broad-Band Random Sigd trons, when he discovered what is now called Gunn effect. This triggered a great deal of research work because the effect A random signal can have some properties which are desir- has applicationsin making transferredelectron-type micro- able at times, e.g., it can be incoherent and broad-band, it can wave oscillatorsand amplifiers. Of course,one should not be used to establish the presence or the direction of location rely onserendipity; there are plenty of other reasons for of.its source, it can simulate a random quantity, and it can be studying noise. This article will present some of them. used to generate another random quantity. These properties Most electrical engineers are aware that random-noise gener- lead to the application of random noise in electronic counter- ators are standard laboratory instruments, so there must be measures, microwave heating by noise, simulation of random quantities,stochastic computing, and generation of random numbers. Manuscript received October 24, 1974;revised March 13,1975. This work was supported by theJoint Services Electronics Rogram. B. As a Test Signal The author is with the Research Laboratory of Electronics and the Department of ElectricalEngineering and Computer Science, Massa- Thereare many instances of measurements in which one chusetts Institute of Technology, Cambridge, Mans. 02 139. needs a broad-band signal with precisely known properties like GUPTA : APPLICATIONS GUPTA: OF NOISE 991 amplitudeprobability density, rmsvalue, orautocorrelation the testing of new radar and sonar systems, and thetraining of function. Electrical random-noise generators are a convenient personnel working with these systems, withoutactually placing source of such signals, having statistical parameters which are the system in field use. This task is carried out by designing either known in advance or can be easily manipulated. This electronic simulators which generate signals resembling those explains the use of noise in such techniques (some of them encountered by the system in actual operating environment. very accurate) as measurement of impulse response, insertion Noise generators are an essential part of such simulators. For loss, and linearity andintermodulation of communication example, in oneradar simulator [3] used for training, the equipment, aswell as in noise-modulated distance-measuring random fading of signals is duplicated by modulating the signal radar. by low-frequency random noise. The modulating signal itself is generated bycross correlating twonarrow-band random- C. As a Probe into Microscopic Phenomena noise signals. In another radar simulator [4], the probability The use of cosmic radio noise to glean information about the of detection of a target is measured by introducing noise into sources of this noise is well known in radio astronomy. Simi- the videodetector. The random characteristics of noise typi- larly, the fact that the electrical noise is caused by the motion cally observed in normal pulsed radar receivers are reproduced (or emission, or recombination, or ionizing collision, etc.) of by adding the output of a noise source to a pulse train and individual carriers suggests the possibility that such micro- rectifying it by a full-wave rectifier before applying it to the scopic processes may be studied through the fluctuations. In video detector. Such measurements of the visual detectability particular, noise measurements can be used for estimating the of signals in the presence of noise are not confined to radar physicalparameters which are related to thesemicroscopic simulation; they are also used in situations like seismic detec- processes. This forms the basis for the application of noise in tion, where the noisysignals are recorded on charts [ 51. the determination of parameters like carrier lifetime in semi- Similar studies of the intelligibility of audio signals in the conductors, fundamental physical constants, and device con- presence of noise are carried out in the evaluation of speech stants, as well as in testing semiconductors for uniformity and communication systems and in physioacoustics [ 61. All such for estimating the reliability of semiconductor devices. measurements may be likened to the measurement of noise figure, in that they determine the performance of a system in D. As a Conceptual Tool the presence of noise. While noise has been the motivating cause for the develop- Noise signals are also used for simulating random vibrations ment of new disciplines like information theory and statistical in mechanicalsystems; the combination of arandom-noise theory of communication, certain other fields, notably circuit generator and a shake table is widely used to test the response theory, have also benefited from the study of noise. In addi- of mechanical