Noise recycling: theoretical analysis and application

Details

Supervisors

Médard, Muriel ; Oestges, Claude ; Vandendorpe, Luc

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil électricien, à finalité spécialisée

Abstract

Noise is a significant challenge in communication systems, degrading performance by reducing the signal-to-noise ratio and increasing error rates. However, traditional approaches based on interleaving may not be optimal in the presence of time-correlated noise. In this context, Noise Recycling has been developed to exploit noise correlation to improve performance. The idea is to recycle the noise by reusing noise estimates to improve signal estimation. This thesis presents a theoretical framework for noise recycling based on Markov chain theory. It deepens our understanding of noise recycling and allows the characterization and computation of error probabilities for both uncoded symbols and error-correcting codes. The model demonstrates exponential convergence of the error probability to a stationary value and provides insight into noise recycling mechanisms, paving the way for method optimizations and extensions. The model is first developed for a general case and then applied to a real telecommunication application, the Single Carrier Orthogonal Frequency Division Multiplexing (SC-OFDM) system after equalization. This recycling philosophy could be applied to any system affected by correlated factors, and in this sense, this thesis also proposes Channel Recycling, a method that uses channel correlation to improve channel estimation.