Echo Cancellation Mode Switching for Digital Connection Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing echo canceller systems fail to timely and properly detect infinite and finite echo return loss (ERL), leading to incorrect echo model generation, signal choppiness, and failure to cancel echoes in digital connections, which can result in the system generating echoes instead of canceling them.
Innovation Solution
The method involves determining a running mean attenuation and echo signal to near-end noise ratio (ENR) to set thresholds for infinite and finite ERL detection, disabling or enabling echo cancellation by bypassing adaptive filters and non-linear processors based on these thresholds and energy distribution analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If echo cancellation is continuously enabled, then echo removal is maintained, but the system generates echoes instead of canceling them in digital connections
Solution Approach 1:
The echo cancellation system dynamically adjusts its operation mode based on detected ERL conditions. When infinite ERL is detected (indicating digital connection), the system switches to bypass mode to disable echo cancellation and prevent echo generation. When finite ERL is detected (indicating analog connection), the system switches to cancellation mode to actively remove echoes. This dynamic adaptation resolves the contradiction by ensuring echo cancellation is only active when physically necessary.
Solution Approach 2:
The system continuously monitors the echo return loss and uses this feedback to determine the appropriate operating mode. The ERL detection mechanism provides real-time information about the connection type, allowing the control logic to switch between cancellation and bypass modes accordingly. This feedback loop ensures the system maintains reliability by adapting to changing network conditions and prevents harmful echo generation in digital connections.
2Reliability
If infinite ERL is not timely detected, then echo cancellation remains active, but the adaptive filter trains on noise and generates incorrect echo models
Solution Approach 1:
The system performs preliminary ERL detection before the adaptive filter can train on noise or generate incorrect models. By continuously monitoring for infinite ERL conditions and switching to bypass mode when detected, the system prevents the harmful training process from occurring in the first place. This preliminary detection action eliminates the need for corrective measures later and ensures immediate protection against echo model degradation.
Solution Approach 2:
When infinite ERL is detected, the system immediately skips the echo cancellation processing path and routes signals through the bypass mode. This rapid switching prevents the adaptive filter from processing noise or incorrect signals, effectively rushing through the detection and response process to avoid harmful training occurrences.
3Reliability
If finite ERL detection is delayed, then echo cancellation cannot be activated, but echo signals are not cancelled and are heard by the far-end talker
Solution Approach 1:
The system maintains continuous ERL detection and monitoring to ensure immediate activation of echo cancellation when finite ERL conditions are detected. The continuous action of detecting and responding to ERL changes ensures that echo cancellation is activated without delay, preventing echo leakage to the far-end talker while maintaining system reliability.
Data Source
AI summary
There is provided a method of detecting an infinite echo return loss (ERL) in an echo cancellation system while in a finite ERL mode. The method comprises determining a running mean attenuation by the echo cancellation system, determining a ratio of an echo signal to a near-end noise ratio (ENR), defining an infinite ERL threshold (THinfinite) as a function of the ENR, and switching to an infinite ERL mode as a function of the running mean attenuation and the THinfinite. The running mean attenuation can be enhanced echo return loss (ERLE), and the higher the ENR the higher the THinfinite and the lower the ENR the lower the THinfinite. The switching can further be a function of an energy distribution, where the switching switches to the infinite ERL mode based on a non-localized energy distribution over an echo path delay for a predetermined period of time.