# What is Signal Processing?

Where do we draw the limits of what is signal processing? Do we include biological, chemical, electrical, mechanical, or other related signals? Do we include non-linear signals? Do we include things that while they might not be thought to be signals in the first place, but might have properties similar to signals?

To facilitate discussion, I'll include a couple excerpts from the classic version of Oppenheim and Schafer:

pg. 6:

A signal can be defined as a function that conveys information, generally about the state or behavior of a physical system...Signals are represented mathematically as functions of one or more independent variables.

pg. 7:

In almost every area of science and technology, signals must be processed to facilitate the extraction of information...These techniques usually take the form of a transformation of a signal into another signal that is in some sense more desirable than the original.

And a couple from Proakis for good measure:

pg. 2:

A signal is defined as any physical quantity that varies with time, space, or any other independent variable or variables. Mathematically, we describe a signal as a function of one or more independent variables.

pg. 3:

A system may be defined as a physical device that performs an operation on a signal...When we pass a signal through a system, as in filtering, we say that we have processed the signal...Such operations are usually referred to as signal processing.

--

Those quotes paint a pretty broad definition of the field. What we have to decide is what we're interested in talking about here. People may not want to talk about, for example, biological systems here, but when it comes to modeling those systems to transform a signal, then it becomes much closer to the domain suggested by the references I cited.

If we open ourselves up to too broad of a definition, it would defeat the purpose of the site, but if we restrict the definition too much, we may never get out of beta, and there won't be a site at all.

In my humble opinion, it's probably reasonable, while we're still in beta, to favor a more permissive definition of signal processing--with the caveat that we try to use apparently off-topic questions to help define where we want to draw the line(s). In other words, rather than completely rejecting the idea of say, a question about the saturation properties of the human eye, we encourage the OP to massage the question until there's consensus that it finally falls just within the boundary of what we're interested in talking about here.

Or, more colorfully, we have to allow (or even encourage) people to paint outside the invisible lines in order to precisely determine where they are. People working in different sub-fields of signal processing will naturally draw those lines in different places, so it's potentially a learning experience for all of us.

• Well said! Funny, I was about to whip out my copy of Oppenheim & Schafer to pull some quotes for this post. We shouldn't aim for anything under the sun just because it is interesting. The site will soon lose focus if we do. However, I disagree that we should let some of these overly broad questions in beta, because beta is when such issues should be hammered out. If we do not resolve this now, it will be much harder to enforce once we're out of beta and have more volume. I agree with the sentiment that massaging a question into being signal processing related would help in borderline cases. – Lorem Ipsum Sep 1 '11 at 13:46
• Right, I'm just saying that we should allow ourselves to spend more time massaging questions while we're in beta, because we'll eventually gain a more precise definition from that process (as opposed to just downvoting without encouraging rephrasing). – datageist Sep 1 '11 at 13:58

I'm going to say that we should include any form of a signal, and any form of processing. A signal can be defined as any measured quantity which varies over time or space, over any number of dimensions. There must be a method of measuring the signal, but the methods could vary significantly.

Examples could include, but are not limited to:

1. Voltages
2. Current
3. Photon count
4. Membrane voltages
5. Amounts of chemicals in a particular area
6. Number of cars passing over a particular patch of road.

Processing could include any of the following, but again, is not limited to:

1. Performing some kind of a filter, linear or non-linear.
2. Looking for frequencies
3. Trying to understand a particular phenomena of a signal.
4. Extrapolating
5. Enhancing
6. Characterizing the original system (IE, finding noise, resolution, etc)

Especially for non-traditional signals, the signals must be doing something interesting in order to qualify. Basically, don't just set them up and let it run (IE, radioactive decay type problem), but rather, the input signal must vary in some method, and the output signal must have some kind of processing done to it that is non-trivial (Not strictly a multiplier or something like that)

• To whoever downvoted, I've added a bit of criteria stating that signals must be non-trivial, which is hard to explain, but easy to spot. A radioactive decay would be trivial, but it might not be trivial if it is periodically being supplied by an external source, for instance. – PearsonArtPhoto Sep 1 '11 at 10:31