Nyquist frequency

Nyquist frequency. In signal processing and information theory, the Nyquist frequency is a critical parameter defined as half the sampling rate of a discrete signal processing system. Named for Harry Nyquist of Bell Labs, this frequency represents the maximum frequency that can be uniquely represented when a continuous signal is converted into a digital signal. The concept is foundational to the Nyquist–Shannon sampling theorem, which establishes the conditions for perfect signal reconstruction from its samples, preventing the distortion phenomenon known as aliasing.

Definition and mathematical formulation

The Nyquist frequency is mathematically defined as half the sampling frequency. If a system samples a signal at a rate denoted by \(f_s\) (in units like Hertz), the Nyquist frequency \(f_N\) is given by \(f_N = f_s / 2\). This relationship is central to digital signal processing theory developed by pioneers like Claude Shannon. The formulation arises directly from analysis in the frequency domain, where the spectrum of a sampled signal becomes periodic. Institutions like the Institute of Electrical and Electronics Engineers standardize these definitions in fields such as telecommunications. The mathematical framework ensures that any sinusoidal component in the original analog signal below this threshold can, in theory, be perfectly recovered.

Relation to sampling theorem

The Nyquist frequency is intrinsically linked to the Nyquist–Shannon sampling theorem, sometimes called the Whittaker–Shannon interpolation formula. This theorem, articulated by Claude Shannon and rooted in work by Harry Nyquist and Edmund Taylor Whittaker, states that a bandlimited signal can be perfectly reconstructed if its highest frequency component is less than the Nyquist frequency. This principle underpins modern audio coding standards like MP3 and AAC developed by the Moving Picture Experts Group. The theorem's proof involves concepts from Fourier analysis and ensures the viability of technologies from compact disc digital audio to medical imaging systems like MRI scanners. Violating this criterion leads to irreversible information loss, a problem addressed in the design of anti-aliasing filters.

Aliasing and folding

When a signal contains frequency components at or above the Nyquist frequency, the phenomenon of aliasing occurs. During sampling, these higher frequencies are "folded" back into the lower baseband spectrum, creating false, lower-frequency artifacts. This effect is described by the Poisson summation formula and is visually analogous to the wagon-wheel effect in film. In practice, systems employ low-pass filters, often called anti-aliasing filters, before the analog-to-digital converter to attenuate these high frequencies. The mirrored spectra are sometimes referred to as Nyquist zones in applications like software-defined radio and radar systems designed by companies like Raytheon Technologies. Understanding folding is crucial for accurate spectrum analysis in instruments from oscilloscopes to spectrum analyzers.

Applications in signal processing

The Nyquist frequency governs the design of nearly all digital systems. In audio engineering, the standard CD sampling rate of 44.1 kHz sets a Nyquist frequency just above 20 kHz, the approximate upper limit of human hearing. Video standards, such as those from the National Television System Committee, also rely on this principle for chroma subsampling. In telecommunications, it dictates the bandwidth of pulse-code modulation systems used in the public switched telephone network. Advanced applications include stochastic resonance in biomedical engineering and the design of digital filters for seismic data processing by organizations like the United States Geological Survey. Space agencies like NASA apply these rules when sampling data from probes like Voyager.

Practical considerations and examples

In real-world systems, the theoretical Nyquist limit is approached with caution. Practical anti-aliasing filters require a transition band, so the usable bandwidth is often less than \(f_s / 2\). For instance, professional audio interfaces from Focusrite or Universal Audio may sample at 96 kHz to provide headroom above the audio frequency range. In software-defined radio using the GNU Radio framework, operators must carefully select sampling rates to avoid aliasing of RF signals. Oversampling techniques, used in delta-sigma modulation for digital-to-analog converters in products from Texas Instruments, effectively increase the Nyquist frequency for improved resolution. Historical systems, like early PCM adapters from Sony, demonstrated these challenges, influencing later standards from the Audio Engineering Society.

Category:Signal processing Category:Information theory Category:Telecommunication theory