Télécom Paris

Estimation of the Signal-To-Interference Ratio Based on Normalized Cross-Correlation with Symmetric Leaky Blocking Matrices in Adaptive Microphone Arrays
  • Thanh Phong Hua, (NEC Corporation/Université de Rennes)
  • Akihiko Sugiyama, (NEC Corporation)
  • Régine Le Bouquin Jeannes, (Université de Rennes)
  • Gérard Faucon, (Université de Rennes)
  • Microphone arrays and array signal processing
  • Transducers and acoustic front-ends
  • Voice activity detection and double-talk detection
Get the paper in PDF format
Acrobat Reader (version 5 minimum) is necessary to read this document.

This paper proposes estimation of the signal-to-interference ratio based on normalized cross-correlation with symmetric leaky blocking matrices for adaptive microphone arrays. Analysis of the normalized cross-correlation (NCC) shows that it is a function of the actual SIR and the phase difference between its inputs. This function gives its optimum SIR estimation when the phase difference is maximized for the interference and minimized for the target.Symmetric leaky blocking matrices are designed and optimized to approximately satisfy this condition. The actual and the theoretical NCCs show good agreement to each other at different SIRs using white signals. Simulation results show that the SIR estimation by the proposed method is close to its optimum behavior, i.e., the NCC equal to 1 for the target and equal to -1 for the interference. Applying the SIR estimation to target detection, a constant threshold can be set for different DOAs and input signals in the proposed method. The maximum estimation error is reduced by more than 34dB to 6dB for a true SIR of 0dB.

©2006 Télécom Paris/TSI
Edition : Télécom Paris -- 2006