What Is Noise?

Noise, in electrical engineering and signal processing, is any unwanted variation in a signal that obscures or distorts the information it carries. It arises from physical processes intrinsic to electronic components, from interference introduced by the environment, and from quantization and rounding errors in digital systems. The study of noise draws from statistical physics, communication theory, and circuit design, and it underpins the analysis of every system that must detect, amplify, or transmit signals.

Characterizing noise requires probabilistic tools because many noise sources are random processes. Engineers describe noise by its power spectral density, its probability distribution, and its correlation structure in time. These statistical descriptions link experimental measurement to theoretical models and enable engineers to predict system performance before hardware is built.

Thermal and Gaussian Noise

Thermal noise, also called Johnson-Nyquist noise, arises from the thermal agitation of charge carriers in any resistive element at temperatures above absolute zero. Its power spectral density is flat across frequency up to terahertz ranges at room temperature, making it the defining example of white noise. Because thermal agitation affects many independent carriers simultaneously, the central limit theorem ensures that thermal noise amplitudes follow a Gaussian (normal) probability distribution. Additive white Gaussian noise (AWGN) is the standard model used in communication theory to benchmark channel capacity and the performance of modulation and coding schemes. The Shannon channel capacity formula expresses maximum data rate as a function of bandwidth and signal-to-noise ratio (SNR), making SNR the central performance metric for virtually every communication system.

1/f Noise and Colored Noise

Low-frequency noise, commonly called 1/f noise or flicker noise, has a power spectral density that increases as frequency decreases, following a roughly inverse relationship with frequency. It is present in transistors, resistors, and semiconductor devices and becomes dominant below frequencies of a few kilohertz in most circuits. The physical origin of 1/f noise is not fully unified: in metal-oxide-semiconductor (MOS) transistors, it is attributed to charge trapping and release at the oxide interface. Colored noise is the broader category for any noise whose spectral density is non-uniform across frequency. Pink noise has equal energy per octave, brown noise has a 1/f² density, and blue noise has energy that rises with frequency. Phase noise specifically describes random fluctuations in the phase of an oscillator signal and is the dominant performance limit in frequency synthesizers and radar systems. NIST's documentation on oscillator phase noise describes measurement and analysis techniques used across precision timing and metrology applications.

Laser Noise

Laser noise encompasses amplitude noise (also called relative intensity noise, or RIN) and phase noise in optical sources. RIN originates from spontaneous emission events that perturb the photon population in the laser cavity, and it limits the dynamic range of optical communication links and optical sensing systems. Linewidth, which is the spectral width of the laser output, reflects the phase noise floor set by spontaneous emission and is quantified by the Schawlow-Townes formula. Semiconductor lasers used in fiber-optic communications are designed to minimize both RIN and linewidth, and low-noise laser standards from NIST's optical frequency metrology group serve as references for optical atomic clocks and precision spectroscopy.

Noise Cancellation

Noise cancellation techniques reduce unwanted noise using signal processing, either by subtracting a reference noise estimate from the measured signal or by destructive interference of acoustic or electromagnetic waves. Adaptive noise cancellation uses a secondary sensor to capture a correlated noise reference and adjusts a filter in real time to track changes in the noise environment. Active noise control in headphones and automotive cabins generates an anti-phase acoustic signal that reduces ambient noise at the listener's ears by 20 to 30 dB at low frequencies.

Applications

Noise has applications as a topic of study and control in a wide range of fields, including:

• Wireless communications: link budget analysis and receiver sensitivity specification using SNR and noise figure
• Medical electronics: low-noise amplifier design for electrocardiography, electroencephalography, and MRI
• Radar and sonar: clutter and interference rejection to improve target detection in low-SNR environments
• Precision instrumentation: noise floor characterization in gravimeters, magnetometers, and atomic clocks
• Audio engineering: signal-to-noise ratio optimization in recording, broadcast, and hearing aid systems