To the Editor:
Stephen Brown, in his May 1997 Lightwave column (page 38), perpetuates a serious technical error regarding the nature of spread-spectrum technology. The following quoted passages are riddled with fundamental errors:
OSpread-spectrum technology is a spin-off of electronic cryptography, which codes a message, subdivides it into several parts, and transmits it in a camouflage of electronic noise by Ospreading? the message over a bandwidth much larger than the message itself. This is an ironic and perhaps a wasteful use [of] bandwidth...O and O ...bandwidth overkill is a virtue in wireless networks. In spread-spectrum transmission, most of the bandwidth is deliberately kept empty to camouflage the message itself. Thus, the information content of a spread-spectrum message will always be far less than what could be achieved in a fiber network. Also, if the demand for bandwidth is expected to grow exponentially in the coming years, then spread-spectrum?s information-carrying capacity will be quickly exhausted.O
The term ObandwidthO is not synonymous with Oinformation-carrying capacity,O as Shannon proved in his revolutionary paper of 1948. The Oband widthO gives only one dimension of a two-dimensional picture. The signal-to-noise ratio (snr) gives the other dimension?the OdepthO of the channel. A narrowband modulation scheme requires a much deeper channel (better snr), while a spread-spectrum modulation scheme requires only a very OshallowO channel (low snr), in order to carry the same amount of information. Shannon proved that the information-carrying capacity of spread-spectrum and narrowband modulations are identical!
Another way to understand this is to look carefully at the carrier signals used by each scheme. A narrowband modulation scheme is based on sine waves, and sine waves of two different frequencies (or wavelengths) are OorthogonalO?i.gif., they have zero cross-correlation. One spread-spectrum scheme is based on pseudo-noise waveforms that are chosen precisely because they are also orthogonal to each other?i.gif., their cross-correlations are nearly zero. If you look at a spectrum analyzer, which arbitrarily divides up the domain into narrowband frequencies, a wideband signal looks like noise. However, if you were to first utilize a despreader, which converts the spread-
spectrum signal into a narrowband signal, then the modulated signal would be narrowband, while any narrowband signal would be OspreadO into a wideband signal! So the underlying theory of narrowband and spread-spectrum systems is identical, once the appropriate family of orthogonal waveforms has been chosen.
It is true that spread-spectrum systems have been used for cryptographic types of communication, but their advantages in wireless communications accrue from more mundane issues such as a greater immunity to multipath fading.
The acknowledged ObibleO of spread-
spectrum technology is Dixon, Robert C., Spread Spectrum Systems with Commercial Applications, 3rd Ed., John Wiley & Sons, New York, 1994, ISBN 0-471-59342-7.
Shannon?s paper can be found in: Shannon, C.gif., OA Mathematical Theory of Communication,O Bell Sys. Tech. J. 27, 3 (July 1948), 379-423, and 27, 4 (Oct. 1948), 623-656.
Henry G. Baker, Ph.D.
Stephen Brown replies:
Dr. Baker is correct?bandwidth alone is not identical to the concept of information-carrying capacity. The second dimension he refers to, snr and associated modulation schemes, represents the ability to subdivide bandwidth into distinct channels of information. For a given bandwidth, the better the subdivision, the greater the information-carrying capacity. Conversely, for a given way to subdivide the bandwidth, the greater the bandwidth, the greater the information-carrying capacity. I will have to rethink the concept of spread-spectrum as sending a message with a bandwidth much smaller the transmission bandwidth, a description carried in several articles I have read. I continue to support my description of spread spectrum being a commercial spin-off of electronic cryptography in a military setting, where much of the original testing and real-world operation first took place. Finally, I continue to support my larger point: the information-carrying capacity of a spread-spectrum RF infrastructure is ultimately limited by bandwidth, a constraint not faced by a fiber system, where many different light frequencies can be applied to the same fiber path to raise the bandwidth capacity by many multiples.