A new and better method of MVDR (minimum variance distortionless response) mode filtering based on mode decomposition

Feng Yi, Chao Sun, Longfeng Xiang, Zongwei Liu

Research output: Contribution to journalArticlepeer-review

Abstract

Mode filters are commonly implemented using the sampled mode shape or pseudoinverse filters, but they depend on the received array length and have sensitivity to white noise. We propose what we believe to be a new and better adaptive mode filter based on the MVDR filtering. We explain it in sections 1, 2 and 3 of the full paper. Their core consists of: (1) the MVDR mode filter can estimate the complex mode coefficients by suppressing other modes and noise; (2) the performance of the MVDR mode filter is analyzed by computer simulation based on the received data and the sound velocity profile in shallow water. The simulation results, presented in Figs. 2, 3 and 4, and their analysis show preliminarily that: (1) the MVDR mode filter achieves the same performance as the reduced rank pseudo-inverse mode filter and offers greater performance than other linear mode filters; (2) the MVDR mode filter can estimate complex mode coefficients more accurately than other linear filters and adapt the filter coefficients to the changing environments using the received data.

Original languageEnglish
Pages (from-to)112-116
Number of pages5
JournalXibei Gongye Daxue Xuebao/Journal of Northwestern Polytechnical University
Volume30
Issue number1
StatePublished - Feb 2012

Keywords

  • Adaptive filtering
  • Algorithms
  • Analysis
  • Arrays
  • Beamforming
  • Decomposition
  • Estimation
  • Frequencies
  • Inverse problems
  • Measurement errors
  • Mode decomposition
  • Mode filter
  • Monte Carlo methods
  • Multipath propagation
  • Sampling
  • Sensitivity analysis
  • Signal processing
  • Signal to noise ratio
  • Simulation
  • Underwater acoustics
  • Velocity
  • White noise

Fingerprint

Dive into the research topics of 'A new and better method of MVDR (minimum variance distortionless response) mode filtering based on mode decomposition'. Together they form a unique fingerprint.

Cite this