{"title":"Introduction to Polyspectral Modeling and Compensation Techniques for Wideband Communications Systems","authors":"Christopher P. Silva, A. Moulthrop, M. Muha","doi":"10.1109/ARFTG.2001.327481","DOIUrl":null,"url":null,"abstract":"The requirements/environments for broadband commercial and military communication systems have focused attention on the issues of efficiency and nonlinearity characterization of power amplifiers, and the mitigation of the distortion they produce. Common system modeling approaches, which are dominated by probing signals of at most a multi-tone nature, prove to be inadequate for representing channels with operational bandwidths reaching the multi-GHz range. This shortfall in modeling fidelity is especially the case for higher-order, non-constant envelope modulations that are often needed to meet bandwidth efficiency demands. As a natural consequence, distortion compensation designs based on these models will likewise prove to be less than optimal in their effectiveness. This paper introduces an established nonlinear system identification technique from the mechanical systems field that essentially solves the wideband modeling problem, and in addition provides very unique assessment and design tools for distortion compensation. The technique is a special case of a formal operator series representation for the given nonlinearity with memory, and its construction is based on an extremely accurate baseband time-domain measurement technique¿using pseudo-randomly modulated signals ¿ that will also be described. The basic modeling and compensation features of the method will then be provided, followed by two representative high-power amplifier (HPA) applications to illustrate and validate the modeling method. The fidelity evaluation will be performed in the time-domain using a normalized mean-square error (NMSE) waveform metric, and compared with results found for standard block model approaches.","PeriodicalId":331830,"journal":{"name":"58th ARFTG Conference Digest","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2001-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"30","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"58th ARFTG Conference Digest","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ARFTG.2001.327481","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 30
Abstract
The requirements/environments for broadband commercial and military communication systems have focused attention on the issues of efficiency and nonlinearity characterization of power amplifiers, and the mitigation of the distortion they produce. Common system modeling approaches, which are dominated by probing signals of at most a multi-tone nature, prove to be inadequate for representing channels with operational bandwidths reaching the multi-GHz range. This shortfall in modeling fidelity is especially the case for higher-order, non-constant envelope modulations that are often needed to meet bandwidth efficiency demands. As a natural consequence, distortion compensation designs based on these models will likewise prove to be less than optimal in their effectiveness. This paper introduces an established nonlinear system identification technique from the mechanical systems field that essentially solves the wideband modeling problem, and in addition provides very unique assessment and design tools for distortion compensation. The technique is a special case of a formal operator series representation for the given nonlinearity with memory, and its construction is based on an extremely accurate baseband time-domain measurement technique¿using pseudo-randomly modulated signals ¿ that will also be described. The basic modeling and compensation features of the method will then be provided, followed by two representative high-power amplifier (HPA) applications to illustrate and validate the modeling method. The fidelity evaluation will be performed in the time-domain using a normalized mean-square error (NMSE) waveform metric, and compared with results found for standard block model approaches.