< retour aux publications

Asynchronous Filters

Auteur(s) : L. Fesquet, B. Bidegaray-Fesquet

Doc. Source: Event-Based Control, Communication and Signal Processing (EBCCSP'2018)

Whatever the systems - large, small, embedded, autonomous… - filters are widely spread and very often mandatory building blocks. However, with the advent of autonomous systems supplied by batteries or harvesters and the popular Internet of Things (IoT) appeared strong power requirements. In order to circumvent this problem, reducing the amount of processed data appears as a good strategy. Indeed, digital equipment already exchange huge data flows and the future promises even more data and extra communications between humans and technological equipment such as robots, cars, planes, etc. Even if there already exists design solutions to enhance the energetic performances of the electronic filters and systems, non-uniform and sparse sampling appears as a relevant solution to drastically reduce the amount of useless data and to go beyond one order of magnitude in terms of power reduction. Indeed, having a reduced number of data limits the computation, the storage capacity, the communications and thus the power consumption. Such an analog-to- digital conversion requires specific skills when designing filters. This talk will present how to address the filter design and synthesis when a non-conventional sampling is applied. This will be illustrated by filter design examples based on level-crossing sampling schemes. Finally, as such sampling techniques are well-suited and compliant with asynchronous logic, some guidelines will be presented for implementing a fully event-driven filter circuit.