Echo Canceller Adaptation Step Control via Expected ERLE
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing echo cancellers face challenges in minimizing convergence time and achieving optimal Echo Return Loss Enhancement (ERLE) for different hybrids in the signal path, particularly under varying line conditions and noise levels, which can result in intrusive echo and poor voice quality.
Innovation Solution
The method evaluates the expected level of achievable ERLE by considering the energy in the reference signal, noise level on the input signal, and estimated Echo Return Loss, and adjusts the adaptation step of the NLMS algorithm based on the comparison between expected and current ERLE, using specific formulas to control the adaptation rate and maximize convergence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If the adaptation step of the NLMS algorithm is increased to reduce convergence time, then the convergence speed improves, but the stability of ERLE maximization deteriorates under varying line conditions
Solution Approach 1:
The patent applies dynamics by making the adaptation step variable rather than fixed. The adaptation step is dynamically adjusted based on the comparison between expected ERLE (calculated from line conditions) and actual ERLE (measured from adaptive coefficients). This allows the system to adapt its convergence behavior to varying line conditions, achieving both fast convergence and stable ERLE maximization.
Solution Approach 2:
The patent implements feedback by continuously monitoring the difference between expected ERLE and actual ERLE, then using this feedback to adjust the adaptation step. The feedback loop compares the desired performance (expected ERLE based on line conditions) with actual performance (current ERLE from coefficients) and modifies the adaptation process accordingly, ensuring both speed and stability.
2Device complexity
If a fixed adaptation step is used in the NLMS algorithm, then the device complexity is reduced, but the ability to achieve optimal ERLE under different hybrid conditions deteriorates
Solution Approach 1:
The patent applies parameter changes by adjusting the adaptation step parameter based on line conditions. Instead of using a fixed adaptation step, the system calculates expected ERLE from line conditions (such as Echo Return Loss) and modifies the adaptation step parameter accordingly. This allows optimal ERLE to be achieved across different hybrid conditions while maintaining reasonable system complexity.
3Reliability
If the adaptation step is aggressively increased to maximize ERLE, then the echo cancellation performance improves, but the convergence stability under noisy conditions deteriorates
Solution Approach 1:
The patent uses dynamics to adjust the adaptation step based on the relationship between expected and actual ERLE. When actual ERLE is below expected ERLE, a larger adaptation step is used to improve convergence speed and performance. When actual ERLE approaches or exceeds expected ERLE, the adaptation step is reduced to maintain stability. This dynamic adjustment prevents aggressive adaptation that would cause instability in noisy conditions.
Solution Approach 2:
The patent implements feedback control by continuously comparing expected ERLE (based on line conditions) with actual ERLE (from adaptive coefficients) and using this comparison to adjust the adaptation step. This feedback mechanism ensures that aggressive adaptation is only applied when it will improve performance, while stability is maintained when the system is already performing well or conditions are noisy.
Data Source
AI summary
A method is set forth for calculating an expected Echo Return Loss Enhancement (ERLE) in an echo canceller. The expected ERLE is used to control the adaptation step of an adaptive filter in the echo canceller. Also, a novel echo canceller is set forth where the adaptation step of its adaptive filter is controlled based on the expected ERLE.