ECAN Metrics › Echo Loss

Echo Loss

Echo return loss (ERL) measures the loss of volume between the original signal and the echo. An ERL of zero is the worst case, meaning that the echo is fully as loud as the original signal. As the delay grows longer, an ERL of up to 55 dB (and at least 15 dB) is necessary to soften the echo enough to avoid distraction. Applying this principle to VoIP, with long delays (up to 150 ms for acceptable call quality), the echo part of the signal needs to be up to 55 dB quieter than the original voice in order to avoid the perception of echo. So high ERL values are good. Note that ERL and related values mentioned in this document are all part of various ITU standards.

The amount of echo suppression is measured in the ERLE, or echo return loss enhancement, which expresses how much quieter the ECAN was able to make the echo, in dB. In other words, ERLE is a measure of what the actual echo canceller is able to accomplish. While ERL measures the “native” echo coming from the tail circuit or far end of the call, ERLE is the amount of additional echo attenuation the ECAN provides. Seen together, they exactly equal the ACOM value.

Another standard metric applied to echo cancellation, the ACOM value is the view of echo from the IP side of the echo canceller. As defined in ITU G.168, ACOM is the “combined” echo return loss through the system, or the attenuation of echo from all possible means.

ACOM resembles ERL. It is a measure of the degree to which an echo signal has been attenuated. The difference is that ACOM is measured on “our side” of the ECAN device. Therefore, because it includes all sources of echo loss in each direction of the circuit, ACOM is the best gauge of echo strength. Like ERL, ACOM should be high.

Figure 1. Echo measurements and echo cancellation.