What Is Signal Sampling?

Signal sampling is the process of converting a continuous-time signal into a discrete-time sequence by recording the signal's amplitude at regularly or irregularly spaced instants. The resulting sequence of values serves as a digital representation of the original analog signal, enabling storage, transmission, and processing by digital systems. Sampling is the essential bridge between the physical world, where signals vary continuously in time and space, and the digital domain, where information is represented as finite sequences of numbers.

The theoretical foundations of signal sampling were established in the mid-twentieth century through the work of Harry Nyquist and Claude Shannon, building on earlier contributions by Whittaker and Kotelnikov. Their results, collectively known as the Nyquist-Shannon sampling theorem, set the fundamental conditions under which a continuous signal can be exactly recovered from its discrete samples.

The Nyquist-Shannon Theorem and Uniform Sampling

The Nyquist-Shannon sampling theorem states that a continuous bandlimited signal containing no frequency components above W hertz can be completely reconstructed from its samples, provided the samples are taken at a rate of at least 2W samples per second. This minimum rate is called the Nyquist rate, and sampling below it causes aliasing, a form of distortion in which high-frequency components are misrepresented as lower-frequency ones in the sampled signal. The theorem was formally stated by Shannon in his 1949 paper, and a classical treatment with historical context appears in the IEEE paper on sampling, data transmission, and the Nyquist rate. Anti-aliasing filters, applied to the analog signal before sampling, suppress frequencies above the Nyquist limit and are a standard component of any practical sampling system. Uniform sampling, where the interval between successive samples is constant, is the most common approach and the one to which the Nyquist theorem directly applies.

Non-Uniform and Compressed Sampling

Non-uniform sampling relaxes the requirement that samples be taken at equal intervals, which is useful when the signal has variable spectral content over time or when hardware constraints prevent perfectly periodic acquisition. Reconstruction from non-uniform samples requires more complex algorithms than the ideal sinc interpolation associated with uniform sampling, and the error characteristics depend strongly on how the sampling instants are distributed. Compressed sensing, developed by Candes, Romberg, Tao, and Donoho in the mid-2000s, demonstrated that signals which are sparse in a known basis, such as most natural images when expressed in a wavelet basis, can be accurately recovered from far fewer measurements than the Nyquist rate would require. The IEEE Transactions on Signal Processing has published foundational and applied compressed sensing research, with applications in MRI, radar, and spectrum sensing. These sampling methods are classified under what practitioners call compressive sampling or random sampling strategies, and the recovery algorithms rely on convex optimization to identify the sparsest consistent solution.

Quantization and Reconstruction

Quantization is the step that follows temporal sampling: it maps the continuous amplitude of each sample to the nearest value in a finite set of representable levels. The number of quantization bits determines the dynamic range of the digital representation, with each additional bit adding approximately 6 dB of dynamic range. Quantization introduces an irreducible error known as quantization noise, which for uniform quantizers with small step sizes approximates a white noise floor. Reconstruction, the inverse process of recovering the continuous signal from its samples, relies on interpolation; in the ideal case governed by the sampling theorem, an ideal sinc filter applied to the sample sequence recovers the original signal exactly. The NIST guidelines on measurement and calibration address how sampling and quantization parameters are specified and validated in precision instrumentation applications.

Applications

Signal sampling has applications across a wide range of disciplines, including:

• Digital audio and music, where compact disc audio uses 44.1 kHz sampling at 16-bit quantization
• Wireless communications, where software-defined radios sample RF signals for digital baseband processing
• Medical imaging, including ultrasound, where spatial sampling determines the resolution of reconstructed images
• Scientific instrumentation, where high-speed analog-to-digital converters sample sensor outputs for waveform capture
• Remote sensing and radar, where pulse sampling and range gating organize the temporal structure of return signals