Saturday, June 15, 2013

Project Ideas for Digital Signal Processing

Dear ALL,

Digital Signal Processing is applied to signals that have no acceptable quality due to the inherent addition of either noise or some sort of distortion they suffer as they pass through a non-linear system.

The processing in the Digital Domain cannot be understood properly without having a complete idea about the type  of distortion that has caused the problem.

So, here is how any DSP solution can be framed :

1. Must have a complete idea about the nature of the input signal  being processed. Theoretical study is necessary here. Almost all the expertise goes into this area.
2. Once the nature of the signal is known and the nature of distortion or noise is know, you will then proceed to create a model which is nothing but a system designed to solve the problem.
3. This is called as the filter design.
4. You already know in how many different ways a filter can be designed.
5. In a mini-project, you have to learn how to start with a problem and end with a model.
6. If you are successful in designing the filter model then last step would be to realise it in C or MATLAB (you should know how to write the code - Practise/learn from examples).
7. You must then select any one of the DSP chips (we have Texas Instruments) must transfer the code into the chip, apply the problem signal and show at the output that the problem is solved.
8. To edit, compile, simulate and debug the code you need an integrated development environment which is different for different chips made by different companies.
9. Code Composer Studio - as the name suggests, helps us in composing the code and it is your studio (place of work) where you develop all your DSP application programs.

The cycle  of writing the code in CCS , dumping into the chip and testing is the easiest part.

The most geek/difficult part is to understand the nature of the input signal and imagine a model/ system/filter that will remove the distortion/noise from the input signal and deliver the clean signal at the output.

Fortunately, in a fixed system and channel everything remains constant like distortion and noise remain same with time.In such cases we can design simple filters.

What if these characteristics CHANGE with time ? This requires a dynamically changing model.
That is where ADAPTIVE FILTERS step in.

Good Wishes...
HoD-ECE-SVIT

No comments:

Post a Comment