No abstract

Abstract-Wideband CDMA systems with orthogonal spreading codes suffer severely due to the loss of orthogonality by multipath propagation. This yields Multiple User Interference (MUI) which gravely reduces the performance of classical systems with Rake-receivers.In our approach we attempt to restore orthogonality again by using a Lattice equaliser. These kind of equalisers are composed of two parts. The prewhitening part and the adaptive equaliser part. The Lattice structure itself has the advantage that its backward prediction errors are orthogonal to each other. Therefore the Lattice structure itself is working as a prewhitening filter. The second part is the adaptive algorithm for the equaliser coefficients. We take the stochastic gradient algorithm for this task and use the pilot channel in UMTS as a reference to calculate the equaliser coefficients. In order to show the feasibility of this approach, some simulation results for channels with low Doppler frequencies are presented and compared with the a system using the classical Rake-receiver and a system with another prefilter approach, the LMS-like Griffith equaliser [1].

Abstract-This paper introduces a Channel EstimaThe basic advantage of this approach is to have good noise reduction capability and good adaptation to the time variant channel coefficients. Some simulation results for one and multiple users for the LIDF-CE compared with other well known CE-filters are presented.

Abstract-Wideband CDMA systems with orthogonal spreading codes suffer severely due to the loss of orthogonality by multi-path propagation. This yields Multiple User Interference (MUI), which gravely reduces the performance of classical systems with Rake-receivers. In our approach we attempt to restore orthogonality by using a modified T-equalizer. Classical T-equalizers are composed as FIR filters with equidistant delays and appropriate coefficients. Their main disadvantage is the high computational effort due to the large number of coefficients.The question arises whether we need to calculate all equally spaced coefficients, or if some of the coefficients can be neglected. The basic idea of this paper is to use only a subset of coefficients and set all others to zero. The main advantages are the reduced computational effort in calculating the filter and the decreased computational costs in using it. In this paper we will show the feasibility with some simulation results for block fading channels. In practice, we usually do not use a T-equalizer very often for time variant channels, due to its high computational costs. An adaptive algorithm is instead taken and modified accordingly. In our case we take a Least Mean Squares (LMS) algorithm for MUI-suppression in a time variant environment. In order to show the feasibility of this approach, some simulation results for channels with low Doppler frequencies are presented and compared with the classical Rake-receiver and a full version of the LMS/Griffith by [1].