This blog will discuss hybrid echo in voice over IP (VOIP) networks and how to hide the effects of echo in the network. Hybrid echo is caused by an impedance mismatch in the hybrid circuit when a two-wire analog phone connection meets a four-wire digital T1 trunk connection.
A certain amount of echo is present in every voice communication, but the echo is not perceived by the listener unless the delay is long and the amplitude of the returned signal is strong enough. Delay is normally measured in milliseconds (ms), while signal strength is measured decibel meters (dBm). Listener echo is more prevalent than talker echo, although both could occur. Listener echo occurs when the transmit voice signal bleeds through to the receive side circuitry.
Echo in voice networks normally occurs in the tail end of a circuit where a Cisco IP phone call is connected to an analog phone at the destination. Echo is a more prevalent in VOIP networks than they are in time division multiplexing (TDM) networks because of the increased delay in voice sampling in voice over IP networks. TDM interfaces sample 8-bit voice samples 8000 times per second (125 microseconds), while VOIP interfaces are sampled every 20 milliseconds (50pps). TDM sampling at 8000 frames per second represents a much lower delay path than 50 packets per second.
Quality of service (QoS) is a requirement for proper voice over IP operation. Improper QoS configuration can lead to echo problems in IP networks, but echo cancellation is leveraged to mask the effects of QoS. Echo cancellation occurs in the DSP hardware that controls the gateway TDM interfaces. Echo cancellation has also been active in Cisco IP phones since the DSP firmware was upgraded in Call Manager 4.0. This post references echo cancellation tuning on gateway TDM interfaces and assumes the voice media (RTP) packets are sent to the priority queue (PQ) QoS mechanism in the Cisco switches and routers.
Echo cancellation is turned on by default, but the echo coverage path may not be tuned to the maximum possible values. Echo cancellation coverage paths are configured in Call Manager with MGCP gateways, while the coverage path is tuned in voice-port configuration mode of H.323 and SIP gateways. The following command should be issued in voice-port configuration mode:
- Router(config-voiceport)#echo-cancel coverage ?
Configure the coverage path to the maximum possible value. 2800 and 3800 ISR routers using PVDM2 modules or Ti5510 DSP in second generation voice hardware use ITU G.168 echo cancellation and can provide up to 64ms of echo coverage page. Dedicated hardware (ECAN) daughter cards can be used on multi-flex trunk (MFT) T1 card modules to provide up to 128ms of echo cancellation coverage path.
The echo return loss (ERL) of the voice interface must also be between the values of 6dB and 20dB for the echo cancellation to activate. If the ERL value is too low, the input gain and output attenuation of the voice interface can be tuned until the ERL value is within the range of 6 and 20dB.
To verify the ERL of an active call, type the command show call active voice | include ERL and view the results. If ERL levels are consistently below 6dB, the ERL worst case coverage path can be configured with the following commands:
- Router(config-voiceport)#echo-cancel erl worst-case 0 (or 3)
To tune the ERL by modifying transmit (gain) power levels and receiver attenuation, a configuration similar to the following can be used:
- Router(config-voiceport)#input gain -2
- Router(config-voiceport)#output attenuation 2
If echo is still prevalent for calls over the gateway interface, the impedance (resistance level) should be configured to match the third party equipment.
Author: Dennis Hartmann