ROBUSTNESS ISSUES OF BLIND CHANNEL IDENTIFICATION/EQUALIZATION ALGORITHMSWITH RESPECT TO CHANNEL UNDERMODELING

Линиите свързват координатора с партньорите.

Накратко на български

Цел на проекта

Research objectives and content Channel equalization is very important for reliable high-speed digital communications. The recent development of second order statistics based blind identification/equalization algorithms has been considered as a major breakthrough and 1 created intense research interest in the area. While many methods can claim exact channel identification/equalization in the noiseless case, under the so-called zero forcing conditions. not much is known about their behavior under less id, conditions, like for example a) additive channel noise, b) channel undermodeling. The robustness of blind identification/cqualization methods with respect to channel noise and/or channel undermodeling is a very critical property, directly related to their applicability in 'real world' scenarios. Our purpose is to study the behavlor of second order statistics or implicitly higher order statistics based blind channel identification/equalization methods (like, for example, the fractionally spaced Constant Modulus algorithm, the Subspace, the Least-squares and the Linear Prediction methods) in the noisy undermodeled cases. The author has studied the Subspace a the Least Squares methods in the noiseless two channel case. The results can be found in A. P. Liavas and P. A. Regalia. ''()n the robustness of least squares and the subspace methods for blind channel identification/equalization with respect to channel undermodeling,''submitted for publication to the IEEE Trans Signal Processing., Nov. 1997. However, further study is required for a lucid understanding of the behavior of the fractionally spaced constant modulus and the linear prediction algorithms. Training content (objective, benefit and expected Impact) The benefits of a deep understanding or the factors that determine the hehavior of blind identification/equalization methods in realistic scenarios are very important from both a theoretical and a practical point of view. Some of them are as follows: a) Anticipation of performance of existing methods under nonideal conditions b) Rigorous comparison of relative performance offerefd by the various methods. c) Isolation of ' weak''points of existing methods and, probably, replacement by more robust ones d) Development of new more robust methods. Links with industry / industrial relevance (22)

Оригинален текст от CORDIS (на английски).

Участници

• NATIONAL AND KAPODISTRIAN UNIVERSITY OF ATHENS · ATHENSКоординаторГърция

Връзки

• Виж в CORDIS