This demo shows the BER performance of linear, decision feedback (DFE), and maximum likelihood sequence estimation (MLSE) equalizers when operating in a static channel with a deep null.  The MLSE equalizer is invoked first with perfect channel knowledge, then with an imperfect, although straightforward, channel estimation algorithm.  The BER results are determined through Monte Carlo simulation.  The demo shows how to use these equalizers seamlessly across multiple blocks of data, where  equalizer state must be maintained between data blocks.

The demo plots the BER results dynamically, per Eb/No value.  While the simulation is running at a particular Eb/No value, the plotted BER point will change its position on the plot to indicate how the collection of many errors reduces the variance of the BER estimate.

The demo also shows the burst error performance of each of the classes of equalizers.  It does so by showing a modified histogram of the intervals between consecutive errors.  If a high percentage of errorless intervals is less than about five bits, then the error performance can be considered bursty.

The following executable files are included with this demo:

eqberdemo - main simulation executive
eqber_adaptive - simulation loop for adaptive linear and DFE equalizers
eqber_mlse - simulation loop for MLSE equalizers with and without perfect channel estimation
eqber_siggen - generation of noisy, channel-filtered BPSK signals
eqber_graphics - functions to create and update data visualization

To run the demo, simply type "eqberdemo" at the MATLAB command line.

In addition, this demo includes the following HTML files that enable you to browse through formatted versions of the simulation code.  These HTML files are:

eqberdemo.html
eqber_adaptive.html
eqber_mlse.html
eqber_siggen.html

The HTML includes images of the demo output at various stages of its execution.