Forrest Sheng Bao's Cyberhome

AVR32 development on Linux

forrest — Fri, 06 Nov 2009 01:44:53 -0500

by Forrest Sheng Bao http://forrest.bao.googlepages.com

I have to say that ATMEL is the most Linux-friendly company among all MCU/DSP makers in the world, followed by TI. Both TI and ATMEL use tons of open source toolchains so Linux zealots can build their apps on Linux.

I learned how to play with AVR 8-bit instruction set when I was an undergraduate. Now, I am gonna use AVR32, which is a 32-bit instruction set.

Here is some good information:
AVR32 32-bit MCU - Tools & Software: http://www.atmel.com/dyn/products/tools.asp?family_id=682#1443
ATEVK1100 development kit: http://www.atmel.com/dyn/products/tools_card.asp?tool_id=4114 only $129

EEG Signal Processing in Python and Scipy.Signal (1): Spectrum Estimation, FIR Filter Design, Convolution and Windowing

forrest — Sat, 24 Oct 2009 20:03:24 -0500

by Forrest Sheng Bao http://fsbao.net

I am doing a take-home midterm test of a class I am taking. So, I decided to use Python to to it. Below is a code for one problem. It includes several frequency used functions in classical signal spectral analysis and FIR filter design. The problem itself is to design bandpass filters over alpha to theta bands and apply them onto a EEG series, and plot the time domain and frequency domain signal, as well as the frequency response of filters. I used window methods to design FIR bandpass filters. The filters coefficients are smoothed by a Kaiser window. Filtering is implemented by convolving original signal with coefficients of filters.

summation and factorial in gprolog

forrest — Fri, 23 Oct 2009 03:47:48 -0500

by Forrest Sheng Bao http://fsbao.net

I can't run sicstus Prolog on my department Solaris workstation. That makes me very unhappy at 3 am. So I decide to use gprolog. And the programming can be easier. In gprolog, FD is built-in.

Factorial in gprolog:

factorial(0,1). 

factorial(N,F) :-  
   N>0, 
   N1 is N-1, 
   factorial(N1,F1), 
   F is N * F1.

Summation in gprolog

sum(0,0). 

sum(N,F) :-  
   N>0, 
   N1 is N-1, 
   sum(N1,F1), 
   F is N + F1.

It seems that "is" is also a built-in predicate in swi-prolog.

STFT using spectrogram() in MATLAB, with Hamming and Chebychev windows and different window lengths

forrest — Wed, 21 Oct 2009 14:59:39 -0500

by Forrest Sheng Bao http://fsbao.net

The code below is an example on using spectrogram function of MATLAB to compute STFT of a chirp signal, which changes frequency from 10 Hz to 200 Hz in 1 second. Both Hamming and Chebychev windows are used. Different window length has been tried, 128, 64 and 32.

Some great books

forrest — Mon, 05 Oct 2009 21:36:48 -0500

by Forrest Sheng Bao http://fsbao.net

There are some great books I always want too read. Following books are so famous that you can find them in online bookstore so easily. I cannot guarantee that you can find them in local bookstores coz they won't be as popular as movie stars' biographies. But they will definitely be in stock in next hundreds of years whereas movie stars' biographies will be dust in next 100 years.

General
The Analects by Confucius and his disciples

Is there a God? Or, are there Gods?
The Varieties of Scientific Experience: A Personal View of the Search for God by Carl Sagan
The Language of God: A Scientist Presents Evidence for Belief by Francis Collins
Things A Computer Scientist Rarely Talks About by Donald E. Knuth

Mathematics

Butterworth, Chebyshev I/II and Elliptic filters in MATLAB

forrest — Wed, 30 Sep 2009 15:49:04 -0500

by Forrest Sheng Bao http://fsbao.net

% Copyleft (C) Forrest Sheng Bao 1984-2009 http://fsbao.net
% with Dept. of Electrical and Computer Engineering and Dept. of Computer Science, Texas Tech University, Lubbock, Texas, USA
%
% This program shows the orders and frequency/phase properties of four different kinds of filters, Butterworth, Chebyshev Type I, Chebyshev Type II and Elliptic.
% The filter is expected to have a cutoff frequency at 200 Hz with attenuation 40 db/octave. The sampling frequency is 2000 Hz.
%
% This program licensed in GNU GENERAL PUBLIC LICENSE (GPL) v3 or later.
% If you do not know what GNU GPL is, please visit http://www.gnu.org/copyleft/gpl.html

Why Z-transform is related to digital filter design

forrest — Tue, 29 Sep 2009 19:23:26 -0500

by Forrest Sheng Bao http://fsbao.net

A friend asked me an interesting question: Why is Z-transform always related to digital filter design? Well, this is interesting. But the clue is simple.

You need a web browser that supports MathML to read this doc. To see whether your browser supports it, please check whether you can see formulas on this page. By default, Linux and Mac user should be able to make it from their Mozilla Firefox browser. For Windows users, well, that's not a question you should ask me.

The author cannot guarantee this article has no typo. Please confirm all equations before you use them.

FT, STFT, DTFT, DFT and FFT, revisited

forrest — Tue, 29 Sep 2009 00:58:15 -0500

by Forrest Sheng Bao http://fsbao.net

The author cannot guarantee this article has no typo. Please confirm all equations before you use them.

There are so many guys asking me the difference between DTFT and DFT. And, they confused these two concepts with FFT and STFT.

Wireless Networking conferences coming

forrest — Sun, 27 Sep 2009 12:34:09 -0500

by Forrest Sheng Bao http://fsbao.net

Conference	Conference time	deadline	Place
ACM MobiSys 2010	June	unknown	Palo Alto, California
IEEE GlobeCom 2010	Dec 6-10	Mar 15, 2010	Miami, Florida
ACM MobiHoc 2010 (w/ MobiCom)	late Sept	Abs: Feb 12, 2010 Paper: Feb 19, 2010	Chicago, Illinois
ACM MobiCom 2010 (w/ MobiHoc)	late Sept	unknown	Chicago, Illinois
ACM SigComm 2010	Aug 30 - Sept 3	Abs: Jan 22, 2010 Paper: Jan 29, 2010	New Delhi, India
IEEE ICCCN 2010	Aug 2 - 5	Abs: Feb 19, 2010 Paper: Feb 26, 2010	ETH Zurich Switzerland

Welch method and autocorrelation of 1-D signals in MATLAB

forrest — Wed, 23 Sep 2009 21:46:22 -0500

by Forrest Sheng Bao http://fsbao.net

Here are some examples about computing autocorrelation, PSD in Welch methods with different windowing functions.

The problems are from John Semmlow's Biosignal and Medical Image Processing chapter 3. I used the MATLAB routine sig_noise() that comes with the textbook CD.

Problem 3.8 Welch method

Problem 3.9 Direct FFT spectrum vs FFT spectrum of a signal's autocorrelation function

Repeat above problem, but let SNR=0 dB