UnderGround Forums
 

AcademicGround >> Critique My C.S. Research Abstract

| Share | Email | Subscribe | Check IPs

2/25/07 6:51 PM
Ignore | Quote | Vote Down | Vote Up
signaljammer
Send Private Message Add Comment To Profile

Edited: 25-Feb-07
Member Since: 06/25/2003
Posts: 2387
 
Please :) Qualitative Evaluation of a True (Physical) Random Number Generator for use in Digital Filter Evaluation. Donald Hersey, Peter Kimmel* Department of Computer Science Digital filters are an important part of the modern technical world. They show up in topics from "least squares" fitting in a physics lab, reducing noise in satellite telecommunications, using census information in sociology research, to analyzing stock markets for predictive trading. It is useful in many fields to go beyond classical filters and design custom Digital Filters for new problems. By using a frequency approach it is possible to completely describe a filter by its "Transfer Curve." Describing this curve by judicious use of Fourier analysis and other methods is fundamental to theoretical filter design. The quality of the implementation of a filter can be described as the correlation of the intended transfer curve and the created transfer curve. Empirical analysis of filters can be done by using a suitably pure noise source as an input and using Fourier analysis on the output of the function to deduce the actual transfer curve. In order to be "suitably pure" a noise source must not correlate to a mathematical (non probabilistic) model. The purpose of my research is to construct a "True Random Number Generator" and determine it's suitability for the empirical analysis of digital filters. The TRNG uses a physical artifact (Sylvania 5722 Diode Tube) to create an electrical potential as a signal. This signal is than digitized using an industry standard analog to digital converter and microcontroller circuit. The Digitized Analog Signal is then outputted as a file onto a standard Secure Digital memory card for later use. This SD card can be read by any personal computer with an appropriate card reader. The file must be checked for randomness before used as an input to the digital filters being analyzed. The TRNG's output will undergo (as defined by Federal Information Processing Standards Publication 140-2 on Security Requirements for Cryptographic Modules) Monobit testing, Poker testing, Runs testing and Long Runs testing. There will also be additional testing to make sure it has a "flat" response to Fourier analysis. This is necessary because it is the deviation from flatness that is used to compute the eigenvalue given by the transfer function for a particular frequency, and thusly deduce the actual transfer curve. It is assumed that a well designed and constructed TRNG will pass the required tests, as the noise artifact in use is a high quality valve noise diode. However, testing is necessary for more reasons than completeness, as the source in question has been out of production for several decades, and aging is a potential factor in loss of entropy in TRNG circuits in general.
2/26/07 12:09 AM
Ignore | Quote | Vote Down | Vote Up
Revolver of Reason
Send Private Message Add Comment To Profile

Edited: 26-Feb-07
Member Since: 01/01/2001
Posts: 38556
I don't see anything wrong with it, but I don't know much about random number use outside of cryptography.
2/26/07 6:27 PM
Ignore | Quote | Vote Down | Vote Up
signaljammer
Send Private Message Add Comment To Profile

Edited: 26-Feb-07
Member Since: 06/25/2003
Posts: 2396
What you mean is a spectally flat noise source (PSD=constant). "Pure noise" does not have to be spectrally flat. True and valid point, after all, pink noise has equal energy by the octave. "Spectrally flat" is exactly what I was trying to convey. I'll have to look up what is specifically ment by zero autocorrelation. My school doesn't have an engineering dept. so I'm taking digital filters as an independant study through the C.S. department. Glad to have some advice on standard phrasing. Thanks both of you :)
2/27/07 10:59 AM
Ignore | Quote | Vote Down | Vote Up
Ted Bennett
302 The total sum of your votes up and votes down Send Private Message Add Comment To Profile

Edited: 27-Feb-07 11:21 AM
Member Since: 01/01/2001
Posts: 5628

I don't know a damn thing about the topic, but I'll venture some grammar ideas :-P

1. In the first line, "show up" is pretty unscientific. I'd lean more towards "appear" or s/t like that.

2. In the second sentence, same part - "they show up in topics from" is wordy, you might be better served by saying "they appear in."

3. "The quality of the implementation of a filter can be described as the correlation of the intended transfer curve and the created transfer curve" - I'm not sure what this is saying, but would it make better sense to say "The quality of the implementation of a filter can be described BY the correlation, etc." ? I don't know, it could be the other way depending on the meaning of what you're saying, which I am not sophisticated enough to know :-)

4. "Empirical analysis of filters can be done by using " - again, not sure how your field works, but that's awful close to passive voice, which is discouraged in most journals I read. Would it sound better to say "Empirical analysis of filters involves using...."

5. "The purpose of my research..." - I was taught on pain of death never to say "I" or "my." ;-) Would it be better to use "this" instead of "my?"

6. "will also be additional testing" - redundant - remove either the "also" or the "additional"

7. "determine it's suitability for the empirical analysis of digital filters" - wrong it's (use "its")

That's off the top of my head, and may be of no use to you whatsoever - my brother is a mechanical engineer, and he and his boys laugh at people who worry about grammar :-P

2/27/07 2:39 PM
Ignore | Quote | Vote Down | Vote Up
signaljammer
Send Private Message Add Comment To Profile

Edited: 27-Feb-07
Member Since: 06/25/2003
Posts: 2398
No, that is great help too!
3/8/07 12:01 PM
Ignore | Quote | Vote Down | Vote Up
iceman420
22 The total sum of your votes up and votes down Send Private Message Add Comment To Profile

Edited: 08-Mar-07
Member Since: 06/30/2003
Posts: 5108
I think a "Zero Autocorrelation Noise Generator" is a better description of what you're trying to build. Well, not a Zero-autocorrelation, but an impulse autocorrelation.

| Share | Email | Subscribe | Check IPs

Reply Post

You must log in to post a reply. Click here to login.