March 2006

AWGN: Still the Great Enabler
By Ed Garcia, President, Noisewave Corp.

If you've looked inside microwave test equipment recently, no doubt you've noticed that there is precious little microwave hardware inside. A great example is the RF power meter, which is now an almost completely digital instrument, except for a little RF hardware in the sensor. In this environment of "RF digitization," a test tool as simple as Additive White Gaussian Noise (AWGN) would seem to be ready for archival in the annals of microwave technology past. Interestingly enough, that's far from the case, as noise-based testing is as viable as ever.

Noise is certainly a signal we are all familiar with, mostly with its deleterious effects. For those new to the positive uses of "white noise," it is a simple but remarkably versatile signal that has a constant spectral density (expressed in watts per hertz of bandwidth) with a Gaussian amplitude distribution. AWGN can be generated by several sources, but avalanche diodes are the most common source used in electronic system testing. When AWGN is injected into the input of a receiver, it can quickly evaluate receiver performance and other parameters. AWGN can be used to simulate a complicated modulation signal, to purposely corrupt an existing signal, as a reference signal in other applications, and as economical source of broadband power.

The Great Enabler
In the 1990s, the availability of coaxial noise sources with precision, traceable outputs allowed some of the first digital wireless systems (such as IS-95 CDMA) to be evaluated in signal environments that were a reasonable approximation of real-world conditions. Today, digitally-generated waveforms have to a large degree replaced AWGN as signal stimuli in cellular and PCS testing, but they simply were not available back when analog systems were transitioning to digital. So in a sense, noise-based simulation enabled the development of digital wireless systems that could withstand dense, hostile signal environments. Today's wireless systems employ complex digital modulation schemes that are evaluated with waveforms and signal densities that are generated digitally and very closely resemble real-world operating conditions. Like so many stories of "modernization," the next sentence in this article would logically read: "digital waveforms have largely replaced diode-base noise sources as signal stimuli in receiver testing." It hasn't happened.

The Leap-Frog Effect
As modulation bandwidths of communications systems increase to accommodate the higher data rates required of high-speed transmission, the systems become more susceptible to noise because more noise power is coupled into the signal. At the highest frequencies and broadest bandwidths, digital waveforms cannot be clocked accurately or even generated at all. The answer: diode-based noise sources, which can have bandwidths greater than 100 GHz and retain their precise nature throughout this range.

As they have before, digital signal generation techniques will ultimately catch up, and noise-based simulation will step aside, its job complete as enabler of another generation of system testing. When the next generation ultra-broadband system appears, noise-based simulation will no doubt be called upon again, followed by its digital alternative, and so on.

Unique Needs For Noise
Noise-based testing is not limited to wireless signal simulation, and has long been a staple in noise power ratio (NPR), bit error rate (BER), carrier-to-noise ratio (C/N), and noise figure measurements. To accurately make these measurements, a precision coaxial noise standard remains essential. For example, noise sources make BER testing practical because the required number of errors can be generated quickly. Without the ability to increase the number and frequency of errors, it could take a week and a half to obtain the number of errors required to determine a system's BER.

Noise sources also make excellent solutions for implementing Built-In Test (BIT), since they are extremely inexpensive yet provide a precise reference by which receiver performance can be measured. Tiny circuit board-mounted noise sources that inject a signal of a precisely-known power level and spectral distribution can be switched-in and the receiver can be checked at various points to determine if it is functioning properly.

Cable model testing presents another application for noise-based testing. Hybrid fiber coax (HFC) systems deliver signals over a broad bandwidth, and are susceptible to noise and adjacent channel distortion. By testing a modem in the presence of noise generated by a noise source, the system can be evaluated and modified if necessary with different levels of filtering and other remedies.

Digitally-generated waveforms may accurately represent signal conditions, but at very high frequencies and very broad bandwidths their effectiveness declines while diode-based noise sources continue to perform satisfactorily. Digitally-generated stimuli repeat, unlike natural phenomena that are problematic at times. In encryption applications this repetition issue can be solved by using a truly random number generator derived from a Gaussian noise source. In BER, NPR, C/N, and other standard receiver tests, noise sources remain a key element of the test system. The amount of information that can be transmitted through a communications channel is a function of the signal-to-noise ratio. The world's persistent hunger for information forces an almost continuous need to evaluate performance as a function of signal-to-noise ratio. In short, noise may not be new and flashy, but it still provides an elegant solution to the problem of evaluating receiver systems of every type for their ability to meet both rated specifications and to glimpse how they will perform in the so called "real world." To paraphrase Monty Python,

"Noise is not dead yet!"

NOISEWAVE
www.noisewave.com
Email this article to a friend!