MVDR Beamforming for Acoustic Interference Cancellation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current beamforming techniques, such as LCMV, face challenges in effectively tracking and canceling undesired noise sources due to sharp directional notches, leading to poor interference cancellation and high computational costs, especially in vehicle hands-free communication systems where driver voice clarity is compromised by ambient noise.
Innovation Solution
A two-stage beamforming approach using Minimum Variance Distortionless Response (MVDR) beamformers with adaptive filtering techniques to isolate and cancel undesired signals, applying linear weights to phase angles and employing adaptive filters to track changes in noise sources, thereby maintaining a distortionless response for the desired signal while minimizing noise power.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LCMV beamforming is used to enhance desired speech signals, then spatial selectivity is improved, but interference cancellation performance deteriorates due to sharp directional notches
Solution Approach 1:
The patent implements dynamic tracking of interference sources using multiple microphones to continuously update beamformer weights. The system adapts to moving interference sources by rapidly recalculating spatial filters, transforming the static LCMV approach into a dynamic system that maintains effectiveness against moving noise sources while avoiding the sharp notch problem.
Solution Approach 2:
The patent modifies the beamforming parameters by using a different optimization criterion that avoids creating sharp directional notches. Instead of the traditional LCMV constraint that minimizes variance subject to a linear constraint, the patent employs an alternative parameter set that achieves interference cancellation without the harmful sharp notches in the spatial response.
2Speed
If frequent updates of LCMV beamformer are performed to track interfering sources, then tracking speed is improved, but computational complexity increases
Solution Approach 1:
The patent divides the beamforming computation into separate independent calculations for each microphone channel. By segmenting the overall computation into parallel channel-processing units, the system can rapidly update beamformer weights without requiring complex full-matrix calculations, thus achieving fast tracking with reduced computational burden.
Solution Approach 2:
The patent replaces the traditional mechanically intensive LCMV optimization calculations with an alternative computational approach that uses simplified weight update formulas. This substitution maintains tracking speed while dramatically reducing the computational complexity and processing requirements.
Data Source
AI summary
The present application relates to a system and method for receiving, via a microphone array, an incoming acoustic signal including a desired signal and an undesired signal, performing a first minimum variance distortionless response beamforming operation on the incoming acoustic signal to isolate the desired signal, performing a second minimum variance distortionless response beamforming operation, by the signal processor, on the incoming acoust signal to isolate the undesired signal, filtering the desired signal to generate a filtered desired signal, combining an inverse of the filtered desired signal with the undesired signal to generate a combined undesired signal, filtering the combined undesired signal to generate a filtered combined undesired signal, combining an inverse of the combined undesired signal and the desired signal to generate an output signal, and generating a data signal in response to the output signal for transmission via a wireless network.