Noise in a communication system can be classified into two broad categories, depending on its source. Noise generated by components within a communication system, such as resistors and solid-state active devices is referred to as internal noise. The second category, external noise, results from sources outside a communication system, including atmospheric, man-made, and extraterrestrial sources.
Atmospheric noise results primarily from spurious radio waves generated by the natural electrical discharges within the atmosphere associated with thunderstorms. It is commonly referred to as static or spherics. Below about 100 MHz, the field strength of such radio waves is inversely proportional to frequency. Atmospheric noise is characterized in the time domain by large-amplitude, short-duration bursts and is one of the prime examples of noise referred to as impulsive. Because of this inverse dependence on frequency, atmospheric noise affects commercial AM broadcast radio, which occupies the frequency range from 540 kHz to 1.6 MHz, more than it affects television and FM radio, which operate in frequency bands above 50 MHz.
A characteristic of electrical communication systems is the presence of uncertainty. This uncertainty is due in part to the inevitable presence in any system of unwanted signal perturbation, broadly referred to as noise, and in part to the unpredictable nature of information itself. Systems analysis in the presence of such uncertainty requires the use of probabilistic techniques.
Noise has been an ever-present problem since the early days of electrical communication, but it was not until the 1940s that probabilistic systems analysis procedures were used to analyze and optimize communication systems operating in its presence. It is also somewhat surprising that the unpredictable nature of information was not widely recognized until the publication of Claude Shannon's mathematical theory of communications in the late 1940s. This work was the beginning of the science of information theory