Stevens REU 2007

Jacob Alperin-Sheriff

Week 1

• 5/20/2007: Drove to Stevens. Arrived at the Stevens Institute of Technology, rooming in Jonas/Technology Hall. Met roommate John Lee.
• 5/21/2007: First meeting with Professor Yao, met other students participating in the REU.  Department Chair Tewksbury gives an inspirational and thought-provoking speech reminding us how far technology has come in recent times, and how cellphones have so much stuff on them and that thanks to them, people can track you. Initial discussion of projects, the signal radio and the prostate cancer (I'm in the signal radio group), but I don't get to discuss my project. Lunch at Karma,  hung out for a little while with other REU students, wandered around Hoboken. Also worked more on my factorization algorithm; the curve created by successive applications of pvals(a,n,k) is logarithmic and may be used to find factors more quickly

    Note: pvals(a,n,k) finds values of p (where pq=n) based on the knowledge that

  p²+(-1-k-lg(k)+(a-n)/k))p + n > 0

• 5/22/2007: Met with Professor Yao to discuss my project. It looks like I will be writing a software backend for a wireless signal radio network.
• 5/23/2007: Read through the Flex-Radio code
• 5/24/2007: Continued reading through the Flex-Radio code. Looked up Chebyshev approximations;

Chebyshev approximations:

A Chebyshev polynomial T_n(z) of the first kind is defined as the following contour integral enclosing the origin, traversed

counterclockwise:               

Using these polynomials, define

c_j =

c_j =

Then

This formula gives a very close approximation to the minimax polynomial, which is important for the purposes of modulation/demodulation.

• 5/25/07-Pizza party; further discussed the Software-defined radio with Professor Yao. Began reading articles describing the basic layout of the SDR software.

Week 2

Learned that my project would involve adapting the Software Defined Radio to support dynamic communication. The goal is to have it change frequencies if the current frequency is “noisy,” that is, already occupied by someone else. Currently, this is not done automatically, but hopefully at the end of these 10 weeks I’ll be able to get it to do so. I fixed the given source code a little so that it compiles and runs. The bugs present are likely not really serious bugs, per se, but just situations that arise when the radio hardware isn’t working correctly with the software. 

I also continued going through and commenting the given code provided by Flex-Radio. I believe I have found the variable that controls the data-rate. However, I’ll need to actually get the radio integrated with the software to be sure.

Background Reading:

Modern Digital Signal Processing by Roberto Cristi

Software Radio by Jeffrey Reed

Goals for Week 3

-Continue reading through and commenting the given code

-Get the radio to work with software so I can actually see what i’s really going on

-Do some testing with the radio to better understand the code

-Continue with the background reading

-Read up on noisy channel conditions

-Use the radio to get a feel for what noisy channel conditions are like

Week 3

• Finally got the radio hardware to work with the software to receive signals. Although I could not get it to actually transmit radio stations like it was supposed to, it was getting significantly stronger signals at the correct frequencies, that is, the frequencies that the radio stations actually transmit at were showing much higher amplitudes.
• Began reading up on methods of converting digital data to an analog signal with algorithms instead of with electronic digital-to-analog converters.

Frequency Shift Keying

I decided to use frequency shift keying to transmit the data, as that was the method used by Nishant Kumar

, a very grammatically challenged former masters student at Stevens, in his modification to the Software Defined Radio several years ago. Although it was tough to figure out what he was saying from his awful grammar and lack of detail, I picked up enough to Wikipedia it.

The way it works is that the frequency changes to signify a 1, and stays the same to signify a 0

Goals for Week 4

• Read up on other algorithms converting digital data to analog signals for transmission purposes.
• Implement a stand-alone test program for frequency shift keying and its variations and possibly for some other DAC algorithms.
• Begin implementing this algorithm into the Software Defined Radio

Week 4

• I finally got the sound to work on the Software Defined Radio. This allowed me to actually listen to various radio stations and see how the words/music translated into sounds.
• I also obtained a microphone and was able to better determine what would give a nice-looking signal with little distortion that I could then modulate to send data over as seen above.
• It unsurprisingly turns out that a simple tone is what the best signal would be. This tone can be raised/lowered small amounts to encode data.
• As part of my being able to modify the signal, I finally began modifying the program to accomplish things. My first modification to the program involved adding a button to the software-defined radio (labeled Jacob). When pressed, the intensity of the signal is lowered by a random amount, different for each period.

Before

After

• I also tried to transmit from one computer to the other. However, for some reason DC power from an adapter is different than DC power from one of those special power things (which, if were an electrical engineer, I'’m sure I’d know the name of; in that case, it would probably also make my project much, much easier to do. 

Goals for Week 5

• Modify the software-defined data to encode data to be transmitted by modulating a tone.
• Continue reading up on new methods of encoding data in a signal.

Week 5

This week I began programming the software defined radio to detect noisy channel conditions. The basis of the

 detection algorithm is standard deviation. I was hoping that if the channel deviates more than a certain number of standard deviations (2.5 seems to be good), then that tells us that the channel is noisy. It works pretty well in most cases. However, if there’s only a gentle curve up from the signal, it doesn’t detect that the channel is noisy even though it seems to be. The checking for the noisy channel is done on the average signal after a few seconds, to minimize the distortions which inevitably occur in radio signals.  Below are some pictures of the signal and the detections.