Glossary term
Signal-to-Noise Ratio
A signal-quality metric comparing desired signal power with noise power over a specified bandwidth and measurement condition.
Definition
metricSignal-to-noise ratio is the ratio between desired signal power and noise power over a specified bandwidth and measurement condition.
SNR describes how clearly a desired signal stands above random or unwanted fluctuations. It is central to communications, instrumentation, radar, audio, imaging, sensors, and digital acquisition because it affects detectability, bit error rate, resolution, estimation uncertainty, and usable dynamic range.
For power quantities, signal-to-noise ratio is:
In decibels:
If voltage or current amplitudes are used across the same impedance, the decibel form becomes 20\log_{10}(A_\text{signal}/A_\text{noise}). The bandwidth matters because integrated noise power usually increases with measurement bandwidth.
Engineering interpretation
High SNR means the signal is easier to detect, decode, estimate, or control. Low SNR increases bit errors, false alarms, uncertain measurements, poor image quality, unstable feature extraction, and reduced effective resolution. Filtering can improve SNR only when the desired signal and noise occupy different frequency bands or have distinguishable structure.
SNR must be stated with its measurement point. Input SNR, output SNR, carrier-to-noise ratio, energy-per-bit to noise-density ratio, and ADC signal-to-noise-and-distortion ratio are related but not interchangeable. A receiver can have good antenna SNR and poor demodulated SNR if noise figure, bandwidth, phase noise, interference, quantization, or clock jitter are poorly controlled.
Measurement context
SNR is not a single universal property of a device. It depends on where the signal is measured, how noise is integrated, whether the signal is continuous or sampled, and whether deterministic distortion is included or excluded. In practical systems, engineers usually separate thermal noise, quantization noise, phase noise, interference, drift, and distortion before deciding whether filtering, shielding, gain redistribution, averaging, coding, or hardware redesign is the right correction.
Common mistakes
A common mistake is quoting an SNR number without bandwidth, weighting filter, detector type, averaging, or reference level. Another is treating all unwanted content as random noise; interference, distortion, drift, aliasing, and quantization error may require different fixes. A strong SNR review states signal definition, noise definition, bandwidth, measurement location, impedance or scaling, averaging method, dB convention, and dominant noise sources.