Approximate signal processing using incremental refinement and deadline-based algorithms

Abstract

A framework for approximate signal processing is introduced which can be used to design novel classes of algorithms for performing D F T and S T F T calculations. In particular, we focus on the derivation of multi-stage incremental refinement algorithms that meet a variety of design criteria on the tradeoff achieved at each stage between solution quality and computational cost.can be used to generate a large number of algorithms whose stages possess different cost versus quality tradeoff characteristics. In this paper, we demonstrate the application of a mathematical framework for quality and cost assessment to the design of incremental D F T refinement algorithms whose tradeoffs meet a variety of design criteria. Since the S T F T can be decomposed into a series of D F T calculations, these design techniques may be applied to the derivation of incremental S T F T refinement algorithms as well.

<title>Mixed-radix approach to incremental DFT refinement</title>

“…The values of X(/c) in equations (8) and (9) differ only in that the DC component of (9) has been lost in the differencing operation of (10).…”

Low Power Digital Filtering Using Adaptive Approximate Processing: Fir Filter Structures

“…Digital filters represent a fundamental signal processing element which is found in all of the portable systems already mentioned and many others. Motivated by the growing demand for low power digital signal processing techniques for use in mobile devices, algorithmic approaches to low power frequency-selective digital filtering have been extensively developed [11][12][13][14][15][16][17][18][19][20]. It has been demonstrated that significant power savings may be achieved in digital filtering applications when the order of a digital filter is dynamically varied to provide time-varying stopband attenuation in proportion to the timevarying signal-to-noise ratio (SNR) of the input signal, while maintaining a fixed SNR at the filter output.…”