Signal-to-Noise Ratio

From Stanford SSI Wiki
(Redirected from SNR)
Jump to navigation Jump to search

Signal-to-noise ratio (SNR) is an important concept in signal processing that describes how difficult a signal is to detect. Signals with high signal-to-noise ratios are easier to detect, as unwanted noise can generally be removed without damaging the clarity of the original signal. Conversely, signals with low signal-to-noise ratios are difficult to detect and require very precise receiving equipment and/or processing techniques.

Signal

A signal, very broadly, is any means by which information is transferred from one location (time/space) to another. All signals have a physical representation, sometimes known as a medium or carrier. Some examples of signal carriers are light, sound, electric current, and magnetic fields.

One of the most important types of signals is time domain signals, signals which are transmitted by something in a medium changing in time. A light blinking on and off and the time-varying sounds produced in human speech are both examples of time domain signals. Radio waves and the electric signals carried over Ethernet lines are also time domain.

Signals are transmitted over a link between two points (or nodes), where one node acts as a transmitter which sends the signal to the other node, acting as a receiver. Bidirectional links are links where the two nodes can change roles, such that signals can be transmitted from either node to the other. All signals are transmitted by the movement of energy from transmitter to receiver through the signal's medium; for example, light waves carry energy from their source to a detector, which is then able to convert that light energy into electric energy. Transmitters output some amount of energy per unit time they are transmitting, so this transmitting power is commonly measured in watts (W).

Noise

Noise is any energy in a communication medium that can be mistaken for an intentionally transmitted signal. For example, in an optical communications link, which consists of a light source (LED, laser, etc.) and a light detector some distance away, any light not from the light source that the detector receives is noise. Other examples include interference picked up on an Ethernet cable or radio antenna. Noise is present to some extent in all real physical systems.

Signal-to-Noise Ratio

As stated above, transmitting a signal requires power, not all of which will arrive at the intended receiver. As with noise, these losses occur to some extent in all real physical systems.

The ratio of signal power that arrives at a receiver to noise power picked up by a receiver is known as the signal-to-noise ratio, which is a unitless quantity that describes how much more powerful a signal is than the noise that could obscure it. Generally, a signal must have an SNR greater than or equal to 1 (meaning that the signal is no less powerful than the noise) in order to be detectable. Higher SNRs allow for faster information transfer and greater reliability of communications.