notor research - What is noise?

Home  Contact Us  FAQs  Links

Noise is interesting stuff.

First of all, noise is really random. By random, I mean unpredictable in its value at any instant in time. 

Noise is an intrinsically statistical entity. It's instantaneous value is not predictable, but its mean, standard deviation (a.k.a. RMS value), and variance (a.k.a. power) are predictable and measurable. 

Everything above a temperature of absolute zero (-273.15 oC) has some noise. In electronics, we are primarily concerned with noise in conductors and semiconductors, but I think it is safe to say that everything in the universe exhibits noisy (i.e. random) behavior of some sort.

One of the favorite kinds of noise in wireless systems is called Additive White Gaussian Noise (AWGN). If you say Additive White Gaussian Noise around a group of wireless engineers, they will be impressed. For example, you could ask "How well does this wireless device work in the presence of additive white gaussian noise?" Then, just stand back and watch the resulting argument.

AWGN is called additive because it, like detergent added to the wash water, is added to a wireless signal as it travels through electronic circuits, the atmosphere, and even outer space.

AWGN is called white because it contains energy at all frequencies with an equal power level at each frequency. 

AWGN is called Gaussian because when you accumulate data on the instantaneous level of the noise signal (instantaneous voltage or current), you find that the instantaneous level has a Gaussian distribution (the well known norma, or bell, curve). 

In fact, the effective value (the DC equivalent of the noise signal) is equal to the standard deviation of the Gaussian distribution. The effective value is also called the RMS (root-mean-square) value. 

And the noise power is equal to the variance of the Gaussian distribution (or the standard deviation squared).

Every electronic conductor in the universe generates thermal noise. The noise is proportional to the physical temperature of the conductor. Thermal noise power is constant everywhere temperature is constant. At 25 oC, the noise power is -174 dBm/Hz. That is, if you somehow measured the noise power generated by a conductor in a 1 Hz bandwidth in a room at 25 oC, you would find that the power is -174 dBm (dB relative to a milliwatt). That is 3.98 x10-21 Watt. 

Thermal noise power increases as the bandwidth you measure it in increases. So for thermal noise in a 10 Hz bandwidth, you would measure 3.98 x10-20 Watt at 25 oC.

You cannot measure noise power this low with real instruments, however. That's because the instruments generate excess noise. The instrument may be warmer than the temperature of the room, so the instrument generates more thermal noise than the conductor at room temperature, masking the noise in the conductor. The instrument has electronic devices whose noise is proportional to the voltage and current which power them. This noise is often much higher than thermal noise.



Home  Contact Us  FAQs  Links

Copyright John Notor, 1998 - 2018, All Rights Reserved.
Last Revised: February 10, 2018