Suréchantillonnage

Upsampling begins by inserting zeros between each sample of a decimated signal. This zero-inserted sequence is then passed through a lowpass filter with a cutoff frequency at the new Nyquist limit. The filter attenuates higher-frequency replicas while retaining original frequency components, producing a higher sampling rate signal that effectively reverses the downsampling process.

Inserting zeros between samples expands the original spectrum and introduces repeated spectral replicas at intervals determined by the new Nyquist frequency. These replicas occur because the zero-insertion process creates a periodic repetition of the signal's frequency content. The lowpass filter subsequently removes these unwanted replicas, leaving only the original frequency components intact.

After zero insertion, a lowpass filter with cutoff frequency at the new Nyquist limit attenuates higher-frequency replicas while preserving original frequency components. This filtering step is essential for reconstruction signal using interpolation principles, ensuring the upsampled signal maintains signal integrity and prevents aliasing distortion in the final output.

Upsampling compresses the Fourier transform toward the origin. For example, a sequence with spectrum spanning from −8π/9 to 8π/9 becomes compressed to −π/9 to π/9 after upsampling by a factor of two. This compression reflects the expansion of the time-domain signal due to zero insertion, inversely scaling the frequency domain representation.

Combining upsampling and downsampling operations allows precise control of sampling rate conversion while preventing aliasing. For instance, upsampling by four and downsampling by nine achieves maximum downsampling without aliasing. This combination effectively adjusts the sampling rate while maintaining signal fidelity and the essential characteristics of the original signal.

Setting the lowpass filter cutoff at the new Nyquist limit ensures that only frequency components within the valid range of the upsampled signal are retained. This prevents aliasing by removing spectral replicas that exceed the new Nyquist frequency. The filter preserves the original signal's frequency content while eliminating artifacts introduced by zero insertion.

Upsampling enables effective management of signal sampling rates in communications, audio engineering, and data compression. By inserting zeros and applying lowpass filtering, engineers adapt signals to different sampling requirements while maintaining signal integrity. This technique balances sampling efficiency with signal fidelity, ensuring accurate signal processing and reconstruction across various technological domains.