Noise causes misery for most engineers. Whether it is internal circuit noise degrading a sensitive transducer signal, EMI/RFI intruding on signals and components, or external noise corrupting incoming wireless signals, we devote lots of time, energy, hardware, coding, algorithms, and software effort to minimize its impact.
Ironically, the ability to deal with noise separates the good engineers from the great and experienced ones. If not for noise, a good design would be much, much easier to do, and a less-skilled engineer could do much of it. But great engineers make many design decisions based on the anticipated or actual presence of noise, whether they articulate this or not. It's as if noise is the ghost that is always hanging around the circuitry and system, ready to strike at any time. Great engineers know you can’t be too careful or complacent, even in the somewhat noise-tolerant digital world.
Great engineers also know how to assess, model, and anticipate noise, and how it will manifest itself, so they can take precautionary steps. Even better, they know how to track it down and squelch it, to the maximum extent possible, to yield a more robust design
Great engineers also understand the varied personalities of signals and noise, and can “become one” with both of these factors. The late Professor William R. Bennett, in his seminal textbook Introduction to Signal Transmission, made this point in the book's introductory chapter. (Full disclosure here: I have forgotten nearly all the complex math and analyses of Chapter 2, onward!)
Bennett points out that there are four permutations of signals and noise: known or unknown signal, combined with known or unknown noise. The easiest class to deal with is know signal and known noise, where many of the characteristics of the signal—its modulation, format, bandwidth, and similar, are fairly well known in advance; and where the characteristics of the noise—such as white noise of a certain bandwidth and rms power—are known. An example would be receiving a AM signal via a wired link, where atmospheric noise is not an issue, only internal circuit noise.
At the other extreme is unknown signal and unknown noise, where very little is known in advance about either. An example would be searching for signals from extraterrestrial beings (the SETI project), despite hard-to-assess noise from galactic, atmospheric, stellar, pulsar, and many other sources. In between the extremes are cases where some things are known about both in advance, but not too much, such as a radar-signal return from an aircraft.
A conscientious designer spends as much time as possible, in the design phase, studying the characteristics of both the signal and the noise. Then he or she can provide the requisite hardware and software algorithms to accommodate, and then master, the signal-to-noise situation, despite challenging circumstances.
Bill Schweber , Site Editor, Planet Analog