The noise bandwidth of sampled data systems

The effect of aliasing on sampled noise power measurements is investigated. It is shown that although aliasing does not bias noise power measurements, it can significantly increase the measurement variance. The equivalent noise bandwidth (EENBW) of a sampled data system is defined by equating the mean and variance of the noise power measurement to those of a system with a rectangular spectral response. With this definition, which is shown to be consistent with that for analog systems, it follows that the ENBW is bounded by the smaller of the analog ENBW or half the sampling frequency. Practical examples of sampled-data systems are used to demonstrate the increased variance, integration time, and noise spectral density that accompany aliasing.>