Interference Mitigation

Vaishnavi worked on Median-based interference identification and removal. The basic idea is to treat the time-freq map of a given baseline as a 2-D array. See the above diagram for such an example, where the red arrow indicates one interference location.

We then take a small section of this 2-D array (say 32x32 matrix), where the 3rd axis is the intensity (or amplitude). We compute the median and standard deviation of the section (32x32 matrix). We then put some criterion of (median+7*sigma) for genuine data.

Amplitude > (median + 7*sigma) is treated as interference. This appears to identify interference quite reasonably. Check the following image, where black pixel indicates interference. Compare that with the top image, where most of the interference is identified.

There is some data loss due to over-correction. Even so, total data flagged is about 15%, which is quite good.

Fractals and image characterisation

Some links are in order

1. Fractal Dimension: Wikipedia
   
2. A course on Fractals in Yale U
3. A course on Fractal dimension from images: Munich U
4. Fractal Dimension explained
   

So, once you know about fractal dimensions, come to read the stuff on the right (Conti, 2001)


One can treat the image 3-D object. Compute the total number of occupied boxes in X-Y-I dimension box, as a function of size of the box. D = ln(number)/ ln(radius).

It is a little bit more complicated. Check the paper by Conci, a PPT talk can also be found.

How to distinguish between landscape and portrait pictures?

1. Perhaps we can search for a large number of pixels with same natural colors: green, blue and black (shadows). look if a large fraction of pixels contain the same 'Hue' and 'Saturation'.
2. Another try: Look at the Fourier spectra of images, and mark radii of 60%, 90%, 99%, 99.9% power. they should be distinct for landscape images and facial portraits or nearby objects.
   
3. Human objects have a lot more symmetry than the natural objects. In fact, there could be some fractal pattern seen over the different length scales of an image of a natural scenery. Try to capture 'fractal' properties of pixels.

speaking of the last one: one could look at fractal dimension of a picture pixel values. How? Perhaps in the next blog post...

BITS Goa Radio Telescope : Software Correlator

Thanks to Aniket and Mandar, we will get a data acquisition card interfaced to a PC by next semester.

All the card does, is to accept AC voltages of 250 KHz bandwidth, digitize it (2-bit ADC) and sample it at Nyquist rate (500 KHz). Four samples (2x4 bits = 1 byte) are then packed together on the fly to form one byte. The resultant one-byte is stored on a PC for processing. So, the pipeline looks as below


--- signal ---- >>-- ADC -->>--Linux PC-->>- FILE
(0.5 V AC________3-level____bit packing
0.25 MHz band)___2-bit_________program


ADC has two comparators (NE 521 ?). Depending upon the input, one of the the following 00 (-2), 01 (-1), 10 (+1), 11 (+2) is the output of the ADC.

The sampler signal of 0.5 MHz samples the ADC output voltages. Four of the samples are fed to the acquisition card through data cables.

The data rates are slow, 500 k Bytes per second. Given the modern computer disk rates, it is possible to sustain a on-the-fly bit packing program in PC. The program accepts 4 bytes, and based on a precalculated table, stores corresponding 1-byte output onto a file.

LO and Sampler Frequency Solution

We have our RF band as 73 - 74.6 MHz. Using one of the frequencies given by the Oven Controlled Crystal Oscillator, we would like to generate LO (of) such (frequency) that, our band is folded at some suitable IF. The end of the IF band should be the sampler frequency, again one of those given by the oscillator.

The first image is one such combination for the RF band and LO (plotted very quickly using PLOT program in Mac).


Given the 0.26 MHz band ending at 4.096 MHz beyond LO (70 MHz), the IF band looks as given in the second figure. The IF band now falls between 0 MHz and 4.096 MHz, although mostly empty (0-3.75 MHz), due to our RF filter.

The band is now sampled with 8.192 MHz, the sub-harmonic of 16.384 MHz from the oscillator.

Since both LO and sampler frequency are derived from the same ref, there should be more stability in the system.

One can have another combination of such a LO and sampler frequencies. As an exercise, try these two frequencies and draw the plots: LO (76.8 MHz) and Sampler (5 MHz).

Oven Controlled Crystal Oscillator MCOCXOW

I looked for crystal controlled COs. Here is one data sheet from Golledge. Excellent stability, we do not know the price. It is their featured product, from price and availability point of view.

fruitful day!

things done:

1. AIPS processing tutorial by Chiranjeevi, it has given a new perspective about GMRT interference removal and calibration.


2. ms for SAX 1808 paper is ready for desh's view.



3. there has been tons of discussion about radio telescope.
   
   
   • the 2-bit samples from 4 elements could combined in one byte per sample rate.
   
   
   • Cross-correlations of each two-element pairs (total 6 pairs, called 'baselines') can be done for N delays. Resultant array of N correlations can be passed over to FFT, resulting in amplitudes and phases for N/2 spectral channels. (This job can be simplified by using tabulated results for all 8-byte combinations).
   
   
   • The value N above is decided by delta-u and delta-v that we can have, using 1/imsize for our map (possibly 1/60 deg).
   
   
   • We need to keep computing such cross-correlations over a large time, integrating amplitude and phases over a large time (about 8 seconds or so).
   
   
   • These integrated amp and phases are the output (dynamic spectrum) of an effective software correlator.
   
   
   
   
   

BITS Goa Radio Telescope

when i spoke about our plans for hardware for our radio telescope, there was some feedback from my supervisor (Deshpande, RRI). he was of the opinion, that

1. our telescope was a fantastic idea, and could eventually do international science using 4 element dipoles. we have to plan and execute it well, of course.


2. in the beginning, we could make images using 2 antennas only, with Mandar and Aniket's data acquisition cards. this means, as soon as we demonstrate 2-slit interferometer, we can think of maps in the second semester.


3. we should think of using 2 bits and less bandwidth. eventually, we will have 4 element dipoles. so, we will have to store 4x2 = 8 bits = 1 byte at nyquist rate. if we opt for smaller bandwidth (0.25 MHz), we could be saving 0.5 MBytes per second on the disk, with each byte being a sample each from 4 elements.




In the immediate plans:



4. we should converge using "Mini Circuits" ICs for our RF components. See their website



5. also, for LO stability, either
   
   

   a) opt for GPS receiver, and use its 10 MHz signal for LO

   
   

   b) use a "oven controlled crystal oscillator"

   
   
   the second option is better for us for now. we will continue to explore the second option for better eventual stability of frequency signal. for better mapping, we need our sampler and LO to be really stable.
   
   

Human Tracking on Campus (update)

Students returned here again. This time they had prepared a lot more for display. They have now decided, as I suggested, to reduce the size of the gadget that everyone would carry with them.

As per the original design, the signal generator gadget sends digital codes as an identification for an individual sender. The transmission is at a given higher frequency, using FM. This would mean that all receiving stations (which are used in triangulation) operate at one frequency, and a person's id is the digital signal sent at that frequency.

Now, we will use individual frequency transmission as an id. we may have to do further tricks, since this might mean very narrow bands and FM using highly accurate filters, etc. one could explore if we could do a crude digital signaling. One could just change the ON/OFF frequency of individual receivers, providing another way of identification...

This circuitry could work close to 65 MHz, providing most of the amplifiers and filters that we need for our radio telescope ;-)

Amateur Radio Telescope at 61 MHz

We finally tested the interference environment around BITS using a TV Yagi antenna. This was at 3 pm on Monday 13th November, 2006.

The frequency region around 60-70 MHz appeared the most calm of all, particularly 61-62 MHz. We need about 500 KHz bandwidth, and we will keep it on 61 MHz. The band at 120-130 MHz appeared to be good as well. It is possible to utilize the same antenna and receiver system at 122 MHz as well, just in case there was intermittent terrestrial source at 61 MHz.

Making A Human Tracker

Students are eager to build a human tracker system in the campus. They want each person to carry a transmitter wherever s/he goes. Multiple towers will receive their signals and convey this reception to the central computer. The computer will time the arrivals with GPS unit, and locate the emitter by simple triangulation. The computer can track the location on a map for an easy display.

Students wanted to know if this was feasible. They had ideas of using high-freq dishes, radio-id tags, etc. However, I suggested them to build a system at low frequencies, which will be less prone to blockage. This is of concern, since the system will otherwise fail if the person reaches in a classroom.

We are not sure about how much of signal is needed to make it detectable at the receiving stations, but we are initially going to play with a system working over less than 300 mtrs, using simple yaggi antennas.

Making an Amateur Radio Telescope

Radio Interferrometer

Today I detailed the entire telescope project with Anita. The standard components will be:

1. Two antennas, separated by about 100 meters. To begin with, of course, we will start with one antenna. Signal frequency of 40/60 MHz +- 5 MHz
   
2. Pre-amplifier will be required in at least one case, to boost the signal by 20 dB and carry it over 100 meters. Signal freq 40/60 MHz +- 5 MHz.
   
3. RF to IF conversion followed by an appropriate filter: 0-2 MHz band chosen.
4. IF amplification, possibly two amplifiers back to back for 40 dB or so gain: signal freq. 0-2 MHz.
5. Phase shifter and adder (with 90 degrees phase difference), followed by a detector. The output is a amplitude time series at the rate of 1/16 second.

This kind of project requires a team of students, possibly 8 students working on different aspects simultaneously. Complex job, but it will be fun if we invest their time in it...