Oversampling

In the digital signal processing is referred to as oversampling and English oversampling [ oʊvɚsæmplɪŋ ], when a signal having a higher sampling rate is processed, as is required for the representation of the signal bandwidth.

An over- sampling of a signal may have applicative advantages. Some of these applications are digital - to-analog conversion, analog - digital conversion, the SC filter

Use of oversampling in the digital-to- analog converters

Oversampling is utilized in the digital-to- analog conversion to

• To remove frequencies above half Zielabtastfrequenz, by overdriving subsequent processing stages to avoid interfering signals,
• To avoid interference by intermodulation this lying outside the transmission band spurious signals in narrow-band signals ( for example, telephones ) represent audible non-harmonic distortion,

The over-sampling is done by a sampling rate conversion from the given Quellabtastfrequenz to the desired Zielabtastfrequenz, which is usually arbitrary and can be suitably selected. Usually, extrapolated to a multiple of the Quellabtastfrequenz, where the factor most often is a power of two.

Use of over-sampling the analog - to-digital converters

Oversampling is utilized in the analog-to- digital conversion to

• Frequencies above half Zielabtastfrequenz be removed ( partial shift from analog filtering in the ( easier to handle ) Digital area) to avoid irreparable aliasing error in the target signal

• This property is required for the sigma-delta - converters

In practice, a signal having a useful bandwidth of, typically, sampled instead. The necessary analog anti-aliasing filters will then instead of a transition zone from the larger transition area of which can be realized much easier.

The oversampling is done through a sampling rate conversion by the most arbitrary Quellabtastfrequenz to the desired Zielabtastfrequenz. This is done by low-pass filtering followed by decimation of the sampling points.

Practice

In practice, the integer frequency oversampling ratios used preferably powers of two. This reduces the computational effort. Oversampling at the higher order required sample rate conversion is often carried out in several stages.

Higher frame rates can be achieved here by the fact that the sum and difference bands are removed at odd multiples of the sampling in the frequency domain. This occurred in the time domain to twice as many samples, the sampling rate is thus doubled. This process is called two times oversampling. For quad - oversampling the sum and difference bands are also removed at even multiples, except in 4 * n, the sampling frequency.

According to the Nyquist-Shannon sampling theorem, the sampling rate must be greater than twice the highest occurring signal frequency so as to allow error-free reconstruction. The theorem assumes ideal anti-alias and reconstruction filters.

In practice, this filter is needed, which have a high slew rate and high attenuation (eg need for a CD player, the filter between 20 kHz and 22.05 kHz fall by about 100 dB). With analog technology filters are not reasonably possible with such requirements. Oversampling, however, allows to move the filter from the analog to the digital domain. The filtering is done with a digital filter at the output only a very simple analog filter is then necessary.

Oversampling does not lead to higher data rates and higher memory consumption. This method is used when reading and not writing data application. A side effect is that is by oversampling the signal to noise ratio, such as CD playback, improved. The noise power is distributed by oversampling evenly over a larger frequency interval.

Reference

Frequently antialiasing is also incorrectly referred to as oversampling.