Sparsity-Aware Adaptive Feedback Cancellation for Hearing Aids
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current adaptive feedback cancellation systems face challenges in effectively addressing sparsity in feedback path impulse responses, leading to slow adaptation and instability, especially in quasi-sparse environments and channels with varying feedback paths.
Innovation Solution
The development of sparsity-aware adaptive feedback cancellation algorithms, such as sparsity promoting LMS (SLMS) and sparsity promoting normalized LMS (SNLMS), which exploit and promote sparsity in the estimated filter responses using diversity measures, allowing for improved adaptation speed and performance across a range of sparsity levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional LMS algorithms are used for feedback cancellation, then the system is simple to implement, but the adaptation speed is slow and steady-state error is high
Solution Approach 1:
The patent changes the parameter p in the diversity measure norm from conventional values to values greater than 2, which fundamentally alters the adaptation behavior of the LMS algorithm. This parameter change enables the system to achieve faster adaptation speed while maintaining stability, directly resolving the contradiction between adaptation speed and algorithm simplicity.
2Stability of the object's composition
If normalized LMS algorithms are used to handle correlated signals, then stability is improved, but high adaptation speed and low steady-state error cannot be maintained simultaneously
Solution Approach 1:
The patent modifies the normalization parameter p in the diversity measure norm to values greater than 2, which changes the balance between stability and adaptation speed. This parameter modification allows the system to maintain both stability and high adaptation speed simultaneously, resolving the contradiction between these two performance aspects.
3Adaptability or versatility
If proportionate NLMS algorithms are used for sparse channels, then sparsity exploitation is improved, but feedback cancellation performance degrades in quasi-sparse channels
Solution Approach 1:
The patent changes the parameter p from 1 (proportionate NLMS) to values greater than 2, which fundamentally alters how the algorithm handles sparsity. This parameter change enables the system to effectively handle quasi-sparse channels while maintaining feedback cancellation performance, resolving the contradiction between sparsity exploitation and cancellation reliability.
4Reliability
If prediction-error-method is used to update pre-filter, then feedback tracking performance is improved, but computational complexity becomes high
Solution Approach 1:
The patent changes the parameter p in the diversity measure norm, which fundamentally alters the computational requirements of the algorithm. This parameter change reduces the computational complexity while maintaining feedback tracking performance, resolving the contradiction between tracking reliability and computational burden.
Data Source
AI summary
A signal processing device comprises: input transducer(s) configured to convert input(s) to an input signal; output transducer(s) configured to convert an output signal to output(s); a signal processing circuit configured to at least subtract a feedback estimation signal from the input signal to produce a feedback compensated signal; and an adaptive feedback estimator. The adaptive feedback estimator comprises processor(s) and machine readable medium(s) collectively comprising instructions configured to cause the processor(s) to: estimate feedback path characteristic(s); construct an adaptive feedback cancellation filter based at least in part on the feedback path characteristic(s); select a value for variable p in a diversity measure norm; compute an update rule for the adaptive feedback cancellation filter, the update rule based on the diversity measure norm; apply the update rule to the adaptive feedback cancellation filter; and generate the feedback estimation signal through employment of the adaptive feedback cancellation filter on the output signal.