Title :
SIMO-NAR NEXT Canceller for DMT DSL-CPE: a Stochastic Shannon Capacity Perspective
Author :
Pierrugues, Laurent ; Duvaut, Patrick ; Pons, Julien ; Santraine, Arnaud
Author_Institution :
Conexant System Inc; ENSEA-ETIS
Abstract :
This paper presents and analyzes a frequency domain near end crosstalk (NEXT) canceller tuned to the DMT-DSL CPE. This per tone NEXT canceller is based on a SIMO (Single Input Multiple Output) structure with a Noise Alone Reference (NAR), strongly correlated to the NEXT disturbance of the main stream and orthogonal to the useful signal. Unlike a time domain algorithm, this two sensors frequency solution meaningfully optimizes the SNR per tone, and thus the aggregate capacity of the DMT-system. Since the NEXT coupling highly depends on the relative position of the victim line and the disturbing sources, we provide an analysis technique that fully takes into account the stochastic nature of the NEXT coupling, similar to outage probability used for wireless communication systems. Based on a new stochastic NEXT coupling model of amplitude and phase, that fits real data measurements, and a Monte-Carlo sampling approach, we observe that the SIMO-NAR solution achieves a minimum of 3dB and 9dB SNR improvement in 50% and 10% of the cases, respectively.
Keywords :
Aggregates; Capacitive sensors; Crosstalk; Frequency domain analysis; Noise cancellation; OFDM modulation; Signal to noise ratio; Stochastic processes; Stochastic resonance; Stochastic systems;
Conference_Titel :
Sarnoff Symposium, 2006 IEEE
Conference_Location :
Princeton, NJ
Print_ISBN :
978-1-4244-0002-7
Electronic_ISBN :
978-1-4244-0003-4
DOI :
10.1109/SARNOF.2006.4534740