Adaptive Crosstalk Cancellation and Lattice-Aided Detection in Multi-user Technologies
Speaker: Mandar Gujrathi, ITEE
When: 2007-08-30 11:00:00
Venue: 78-622
Host: Vaughan Clarkson
Abstract:A multi-user communication system can be highly susceptible to time
variations in the channel, giving rise to crosstalk in the system
which adversely affects the bit-error rate, data rate and/or service
reach. Classical linear crosstalk cancellation techniques, such as
LMS, sometimes fail to adapt quickly enough to the time
variations. In the first part of this talk we propose a new adaptive
partial crosstalk canceller with improved convergence. To achieve
this, our detection-guided algorithm takes advantage of the
sparseness in the channel crosstalk matrix and the relative strength
of the direct channels that can be assumed in certain multi-user
environments, e.g., digital subscriber lines.
In the second half of the talk, we consider maximum-likelihood (ML)
techniques for user detection. ML detection (sphere decoding) has
optimal performance but involves high complexity. We adapt for
multi-user detection the replica technique originally used to study
the thermodynamics of disordered systems in physics. We analyse our
new 'replica algorithm' and demonstrate its performance advantages
over competing detectors.
Biography:Mandar is a PhD student in the Division of Electromagnetics and
Imaging within the School of ITEE.
Type: Ph.D confirmation
Contact:Vaughan Clarkson, seminar host (v.clarkson@uq.edu.au)
or Guido Governatori (ITEE seminar co-ordinator)
(guido@itee.uq.edu.au)