What Is Signal Detection?

Signal detection is the discipline concerned with determining whether a signal of interest is present in a noisy observation and, when it is, with extracting relevant parameters such as amplitude, timing, or direction. The problem is inherently statistical: observations contain random noise, so any decision rule will sometimes declare a signal absent when it is present (a miss) or declare it present when it is not (a false alarm). Signal detection theory provides the mathematical framework for characterizing and optimizing these tradeoffs.

Originating in radar and sonar research during the mid-twentieth century, signal detection methods now appear in wireless communications, medical imaging, audio processing, and optical networking. The central tool is hypothesis testing: one hypothesis asserts noise alone, and the competing hypothesis asserts signal plus noise.

Hypothesis Testing and Detection Fundamentals

The Neyman-Pearson lemma establishes that the likelihood ratio test maximizes the probability of detection for a fixed probability of false alarm. This result justifies the matched filter, which correlates the received waveform against a known signal template, as the optimal linear detector when the noise is white and Gaussian. When the signal is unknown or the noise statistics are uncertain, more robust approaches such as generalized likelihood ratio tests replace exact likelihoods with estimated versions.

Performance is summarized by the receiver operating characteristic (ROC) curve, which plots detection probability against false alarm probability as the decision threshold varies. IEEE Xplore publications on detection theory cover classical results and recent extensions to non-Gaussian noise models, distributed sensor networks, and adversarial settings.

Constant False Alarm Rate Detection

In radar and sonar, the background noise level varies across range and angle due to clutter from terrain, sea surface, and weather. Constant false alarm rate (CFAR) detectors adapt the decision threshold to local noise estimates so that the false alarm rate remains approximately constant regardless of clutter intensity. Cell-averaging CFAR computes a threshold from the mean of reference cells surrounding the test cell; variants such as ordered-statistics CFAR are more robust when some reference cells contain interfering targets.

NIST's work on statistical decision procedures provides foundational references for the statistical estimation methods that underlie adaptive threshold setting.

Multiuser Detection

In code-division multiple access (CDMA) and other shared-channel systems, signals from multiple users arrive simultaneously at the receiver. Conventional matched filtering treats interference from other users as noise, leading to degraded performance at high loads. Multiuser detection treats all users jointly, using the known structure of their spreading codes to separate them. Optimal multiuser detectors achieve significant capacity gains over single-user receivers but at substantial computational cost; suboptimal approaches such as successive interference cancellation and linear MMSE filters balance performance and complexity.

Acoustic and Optical Signal Detection

Acoustic signal detection addresses the recovery of sound sources in reverberant or noisy environments. Microphone arrays and beamforming algorithms improve directional sensitivity, while spectral subtraction and Wiener filtering suppress broadband noise. Applications range from speech enhancement in hearing aids to passive sonar for underwater target detection.

Optical signal detection involves recovering information from light, whether in fiber-optic communications, free-space laser links, or fluorescence microscopy. Coherent detection, which mixes the received light with a local oscillator before photodetection, recovers phase and amplitude and enables advanced modulation formats. Research on optical detection sensitivity limits appears regularly in Nature Photonics.

Applications

Signal detection methods serve a wide range of fields:

• Radar and sonar: CFAR detectors locate targets in clutter-contaminated environments for air traffic control and submarine surveillance.
• Wireless communications: multiuser and MIMO detectors separate simultaneous transmissions in 4G and 5G base stations.
• Medical imaging: ultrasound and MRI systems use matched filtering and statistical detection to identify tissue boundaries and lesions.
• Optical communications: coherent receivers detect phase-modulated optical signals at near-quantum-limited sensitivity.
• Cognitive radio: spectrum sensing detects the presence of primary-user signals so secondary users can access unoccupied channels without causing interference.