When designing the GUI, I had to take in to account who will be using the software. Since the software will be available on the web, the GUI must be easy enough to understand and be logically arranged for users not familiar with all aspects of multiuser detection. The software lends itself to be divided into two parts: the received data generation and the application of multiuser detectors. The design consists of a main control window from which you can set the system parameters to generate received data or test a linear detector. The system parameters are set in their own control window and each detector has its own corresponding control window. This makes the program very modular and easy to follow.
Since there are so many variables to set in each control window, it is easy to chose a value for a parameter which is not valid, resulting in bizarre errors. I employed two different ways to overcome this problem. The first was to make pulldown menus for all parameters which have a fixed number of values. For instance, when you choose Gold codes as the pn sequence, there are only 2 valid options for the spreading gain. These two values are represented in a pulldown menu. However, if you choose random spreading codes, the spreading gain can be any value. In this case, the spreading gain pulldown menu becomes an edit menu, allowing any value to be entered. I also incorporated this idea into the individual detector control windows in the form of pulldown menus for the user and delay parameters. These menus are updated each time new received data is generated.
The other technique I used to overcome erroneous errors caused by invalid parameter selection was to create a message window. When invalid parameters are chosen, the message window makes you aware of this fact and explains why the value is invalid. For example, if the number of in-cell users is chosen to be larger than the spreading gain, a message appears in message window tell you of this fact.