PDF of this tutorialIn modern telephone networks, echo cancellers are typically positioned in the digital circuit, as shown in Figure 4. The process of canceling echo involves two steps. First, as the call is set up, the echo canceller employs a digital adaptive filter to set up a model or characterization of the voice signal and echo passing through the echo canceller. As a voice path passes back through the cancellation system, the echo canceller compares the signal and the model to cancel existing echo dynamically. This process removes more than 80 to 90 percent of the echo across the network. The second process utilizes a non-linear processor (NLP) to eliminate the remaining residual echo by attenuating the signal below the noise floor.
Figure 4. Typical Location of Echo Cancellers
Today's digital cellular network technologies, namely TDMA, CDMA, and GSM, require significantly more processing power to transmit signal paths through the channels. As these technologies become even more sophisticated, echo control will be more complex. Echo cancellers designed with standard digital signal processors (DSPs), which share processing time in a circuit within a channel or across channels, provide a maximum of only 128 ms of cancellation and are unable to cancel acoustic echo. With network delays occurring in excess of 160 ms in today's mixed-signal network infrastructures, a more powerful, application-specific echo-cancellation technology is required to control echo across wireless networks effectively.
