Approximate Eigenstructure of LTV Channels with Compactly Supported Spreading
Abstract
Description
In this article we obtain estimates on the approximate eigenstructure of channels with a spreading function supported only on a set of finite measure $|U|$.Because in typical application like wireless communication the spreading function is a random process corresponding to a random Hilbert--Schmidt channel operator $\BH$ we measure this approximation in terms of the ratio of the $p$--norm of the deviation from variants of the Weyl symbol calculus to the $a$--norm of the spreading function itself. This generalizes recent results obtained for the case $p=2$ and $a=1$. We provide a general approach to this topic and consider then operators with $|U|<\infty$ in more detail. We show the relation to pulse shaping and weighted norms of ambiguity functions. Finally we derive several necessary conditions on $|U|$, such that the approximation error is below certain levels.
5 pages, 1 figure, submitted to the 2007 IEEE International Symposium on Information Theory; condition in Lemma 6 and constants in (26) corrected
5 pages, 1 figure, submitted to the 2007 IEEE International Symposium on Information Theory; condition in Lemma 6 and constants in (26) corrected