Introduction:

We are doing this project, because we are interested in DSP and we will gain valuable experience from this project. The project also suits our background using Matlab and programming in the assembly language. This project will further our background in signal processing and communications. Furthermore, we are excited about working with adaptive filters since this material is a continuance of what we have learned from our digital signal processing classes.

We propose to design and implement an echo-canceling device. We will use a Motorola 56000 series DSP chip, to simulate our echo and implement the filters necessary. The most important filter is the adaptive filter. Adaptive filter coefficients change with time and are a key part in echo cancellation. We will start by testing our theory and filters in Matlab. After we have finished our simulation we will port our algorithm into the DSP chip. Next, we will test our system first by sending a sinusoid through, followed by a more complicated signal.

A significant problem in communications is the generation of echoes. The echoes arise for a number of reasons, with the primary reason being an impedance mismatch. The impedance mismatch occurs when the two-wire meets the four-wire connection, also known as the hybrid. This impedance mismatch causes some of the energy to be returned to the source, as echo.

The delays between primary and echo signals are directly related to the transmission distance. For example, if were to send a signal to a satellite that redirected the signal back to another location on earth, that signal would have a very large time delay compared to a signal sent to a local switching station and back. Short delays will not affect the quality of the signal as much as longer delays. By short delays, we are referring to delays of less than 50ms. Delays of this length are not noticed by the person on the receiver end and are therefore not considered an annoyance.
 
 

Design Considerations:

The objective of the adaptive filter is to minimize the mean-square error between the output of the system and that of the adaptive filter. The adaptive coefficients are an estimate the echo. The hybrid system may be nonlinear, but we will model it with a transfer function with attenuation and time delay (phase shift). We believe that since the filter is time varying the adaptive filter will mimic the nonlinear effects in more complex signals.

The echo cancellation device is shown in figure 1. We chose the input signal s(t) to be a sinusoidal input at 2.5Khz because voice information exist from 0-4KHz. Our filter H(z) is the model of the hybrid system that causes the echo.

The filter has a transfer function of H(z) = .5 z^-1.

The filter will output the echo signal.

This echo signal will then be added to two places.

1. It will be added to the original signal to simulate a signal plus echo.
2. It will be sent to the adaptive filter output for comparison.

The adaptive filter will use these two signals to determine the error.

The adaptive filter will then subtract echo by using its estimate of the echo signal.

As the number of iterations increase the error will decrease.

Thus, the adaptive filter will adapt to cancel out the echo..

Figure 1

We may also try to implement other adaptive algorithm for comparison to the LMS algorithm. Such as BLMS, NLMS, etc.
 
 

Specification:

We want fast convergence to zero in approximately 20 iterations. We also hope to get a good suppression of the echo signal. We will also try to match the Matlab simulation as our specifications for the real-time implementation. We will try to show that the output of the adaptive filter system signal looks very close to the actual signal.

We will have Matlab plots showing the time domain and frequency representation of our signals. We will also show plots of our filters ability to reduce the error of the incoming signal, with respect to the number of iterations that it performs. We will demonstrate that out system works using a several channel oscilloscope to show our inputs and outputs.
 
 

Timetable

Feb 10 Research - Design Blocks

Feb 17 Initial Proposal (BOTH)

Feb 17-24 Refine Proposal (BOTH)

March 22 Simulations in Matlab done (BOTH)

April 5 Assembly Code written (BOTH)

April 12 Test Measurement Done (BOTH)

April 19 Paper Done (BOTH)

April 27 Presentation

May 4 DEMO !!!!
 
 

Parts & Cost

Motorola 56002 DSP MODULE $149

Motorola 56002 EVAL KIT $1500

Matlab Software $400

Labor = 2.5*Hrs*Salary/Hrs