Home         Authors   Papers   Year of conference   Themes   Organizations        To MES conference

Criteria for the numerical evaluation of data recovery algorithms for analogue-information converters
Authors
	Bykova A.V.
	Polunin M.N.
Date of publication
	2020
DOI
	10.31114/2078-7707-2020-3-224-229
Abstract
	traditional approaches to signal sampling in modern systems with high bandwidth lead to a number of problems. Alternative approaches to signal sampling, such as the theory of compressed sampling, are used to solve these problems. Devices based on the principles of compressed sampling are called analog-to-information converters (AICs). Using AIC allows you to reduce the sampling frequency of ADC while maintaining the available frequency range, which leads to lower power consumption and simplification of the ADC design. Structurally, analog-information converters consist of three parts: a reading system, an ADC and a data recovery system. The problem of efficient reconstruction of a signal from compressed data plays an important role in the theory of compressed sampling. There are a large number of different algorithms for recovering a signal from compressed data, but most of these algorithms have great computational complexity. In connection with these, in almost all currently existing devices, the task of signal recovery is delegated to the autonomous processing of the CPU or GPU. The variety of approaches to the problem of signal recovery has generated a large number of algorithms, each of which has its own advantages and disadvantages. In this regard, the task of choosing an appropriate algorithm becomes a daunting task for an AIC developer. The choice of algorithm will depend on various factors, such as the architecture of the sampling system, the requirements for the final device, and so on. In order to simplify the task of choosing an algorithm, this article proposes an estimated coefficient that allows you to compare different data recovery algorithms. The coefficient includes the following criteria: relative estimation of occupied memory, which is important when integrating a data recovery system in a VLSI, the number of iterations and the permissible error. In addition, the article compares various greedy algorithms using the proposed estimated coefficient.
Keywords
	theory of compressed sampling, compressive sensing, analog-to-information converters, data recovery system, data recovery algorithms, greedy algorithms.
Library reference
	Bykova A.V., Polunin M.N. Criteria for the numerical evaluation of data recovery algorithms for analogue-information converters // Problems of Perspective Micro- and Nanoelectronic Systems Development - 2020. Issue 3. P. 224-229. doi:10.31114/2078-7707-2020-3-224-229
URL of paper
	http://www.mes-conference.ru/data/year2020/pdf/D084.pdf