Premium
Time‐variant equalization using a novel non‐linear adaptive structure
Author(s) -
Cowan C. F. N.,
Semnani S.
Publication year - 1998
Publication title -
international journal of adaptive control and signal processing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.73
H-Index - 66
eISSN - 1099-1115
pISSN - 0890-6327
DOI - 10.1002/(sici)1099-1115(199803)12:2<195::aid-acs487>3.0.co;2-k
Subject(s) - computer science , fading , a priori and a posteriori , channel (broadcasting) , equalization (audio) , lti system theory , adaptive equalizer , sort , linear prediction , adaptive filter , algorithm , linear system , control theory (sociology) , mathematics , telecommunications , artificial intelligence , mathematical analysis , philosophy , control (management) , epistemology , information retrieval
In much of the work which has been performed in the area of adaptive equalization over the last 30 years there has been a concentration of effort on systems which are a priori unknown but time invariant (or, at least only slowly variant with time). In certain key applications, most notably in digital mobile communications, this assumption of time invariance is not realistic. It is therefore important to focus on the particular requirements placed on adaptive equalizers in this sort of challenging environment. This paper begins by examining the limitations of classical linear equalizers in the time‐variant environment. A novel, non‐linear, filter architecture which is designed to take advantage of the time‐variant channel is then proposed. Two key points relating to this structure are that it has a complexity comparable to the standard linear forms and it only requires to be adapted using a basic stochastic gradient algorithm. Computer simulation studies are presented which demonstrate the ability of this structure to produce enhanced performance (in terms of bit error rate) across a range of fast‐fading channel conditions when compared to standard linear equalizers. © 1998 John Wiley & Sons, Ltd.