Engineers involved with signal processing are familiar with two concepts regarding the required sample rate (the so-called Nyquist rate) and the 6 dB-per-bit rule. For a given system having a signal bandwidth of B Hz, the Nyquist rate requires that the signal be sampled at a minimum of 2 x B samples per second. The 6 dB-per-bit rule states that each bit in an ADC or DAC converter sample represents 6.02 dB of dynamic range (DR).

 

Given these two well-known rules, this TechNote discusses the conditions under which a signal can be represented using fewer than 2 samples per Hertz, and less than 1 bit per 6 dB of dynamic range, in a way that delivers equivalent results for DSP applications. The article describes how designers that size their system bandwidth and storage requirements based on the Nyquist rate and the 6-dB-per-bit rule can do significantly better -- and, surprisingly, with equivalent results -- by using real-time compression. The TechNote provides examples of real-world signals that illustrate a good enough principle using objective system metrics, where compression reduces the system data rate below the bit rate specified by the Nyquist criteria and the 6 dB-per-bit rule.

 

Readers will come away with a new appreciation regarding the minimum bit rate necessary to adequately represent a sampled signal that meets measurable, end-to-end system requirements. Real-time compression allows DSP users to efficiently reduce the bandwidth predicted by the Nyquist criteria and the 6-dB-per-bit rule.