w00t!

[Technomancer]

I just got an article accepted for publication. It'll appear in a special issue of the "International Journal of Engineering Education".

Title:
"A Simulink Laboratory Package for Teaching Adaptive Filtering Concepts"

Abstract:
This paper describes a Simulink laboratory package for teaching adaptive signal processing concepts. Each lab is designed to convey certain important features of a particular adaptive filter, and to provide comparisons with similar adaptive filtering algorithms. The filters covered include the LMS, nLMS, RLS and GAL, as well as three members of the Kalman filter family and the particle filter. In addition to learning adaptive filtering concepts, these labs also help to ensure a greater familiarity with Simulink on the part of the student.

[shooraijin]

Wow ... I just wish I knew what that all was ...

Care to give us a quick intro?

[Mithrandir]

Erai desu nee.

You did a good jorrrrb.

[Technomancer]

Wow ... I just wish I knew what that all was ...

Care to give us a quick intro?

Sure. Basically, the way a lot of students are introduced to signal processing/telecommunications is by allowing certain assumptions about the signals or the channel. Namely, that the signal's statistics are already known, as is the nature of the channel (and that both do not change over time- this is called stationarity). In many real-life problems this doesn't hold, especially in the more interesting, cutting edge applications. In adaptive signal processing, we weaken the assumptions about prior knowledge and system stationarity. The remaining information is filled in by the signal itself, if we can extract the relevant information. In other words, the processing system adapts to the signal or to the environment (or both), so that we can achieve our principle objective, which is to use the information being transmitted. And since the channel/signal changes often with time, this adaptation should occur over time as well.

What we interested in doing (my co-author and I) was to teach some of these concepts using the Simulink software package. This is a MATLAB-based graphical system design package that is widely used in industry for control systems, telecommuncations, etc. The reason for doing this was two-fold: to introduce the signal processing concepts, and to aid the students in acquiring experience with an important design tool. Using Simulink, the students can abstract away many of the programming details and just come to grips with the algorithm's structure and capabilities. In addition, since Simulink uses hierachical blocks, this also allows the students
to make use of the algorithms in a real-world applications, and to combine them where appropriate. For example, one algorithm is better suited to speech compression, while another is better suited to digital equalization. The two algorithms can be easily combined in a simple application that encodes, modulates, and transmits the data and another that handles similar functions for a receiver. And as I mentioned, since the package is graphical, it is easy to see how the components all fit together.