Fractional-Octave Filters for Energy-Efficient Noise Measurement for Sensor Networks

This paper deals with design of digital bandpass filters for fixed-point arithmetic that fulfills requirements of IEC 61260–1 standard for fractional-octave-band filters. The filters use coupled form and downsampling that provides good approximation of the relative attenuation for fractional-octave-band filters and good stability at low-power fixed-point microprocessors which are essential for energy-efficient processing in sensor networks. For the 16-bit resolution, the IEC 61260–1 Class 1 requirements are fulfilled for designed octave-band filters from 63 Hz and for third-octave-band filters from 50 Hz. Implementation of the IEC 61260–1 fractional-octave-band filters in low-power noise sensors enables advanced measurements that conform international standards for noise measurements in the smart city or in other sensor network infrastructures.