next up previous
Next: The GALFA Surveys Up: Minutes of the 1st Previous: The WAPP

Minutes of the talk on:
      RFI Issues at AO: Present and Future

Murray Lewis (NAIC)

A more detailed version of this talk by the speaker is available in HTML, PDF or PPT format. See also the NAIC spectrum management page.

Note that the full ALFA band covers 1225 to 1525 MHz. Current allocations in the frequency range around that of ALFA include:
960-1215 MHz Aeronautical radio navigation
1300-1350 GPS, Gallileo
1350-1390 Try to avoid, but can be used
1395-1400 Satellite uplink desired
1400-1427 Radio Astronomy
1427- Mobile, aeronautical stuff

The Global Positioning System dubbed GPS is a constellation of satellites of which some are always above the horizon (5-8 satellites always available) so it's always visible. The three principle transmitter bands are denoted: L1, L2 and L5

General coverage: -158 dB. Note that:

Galileo:    2nd generation GPS, under development The use of C band (5 GHz) is under consideration The current design for the Galileo constellation is: The current Galileo schedule is:

RFI at Arecibo:

NAIC maintains an RFI Web page. See also Phil's page.

At Arecibo, RFI is monitored by an omnidirectional hilltop antenna (1200-1450 MHz).

News from this week is that the SJU radars may disappear because USN may withdraw from their base on the eastern end of the island, Roosevelt Roads, once the army gives up its bombing range on Vieques. These radars are at 1350MHz at 50-60 dB above the system noise.

The Puerto Rican National Guard also maintains a radar at 1271 & 1291 MHz; also Puntas Salinas

In the surroundings of the currently protected 1400-1427 band. US has proposed a satellite system with an uplink around 1400 MHz, and a downlink around 1427 MHz. Note that the timing explorer on Space Lab was at 1427 MHz.

Note that the web site provides graphical records of the fractional occupation of the spectrum by rfi in 0.1 MHz bins, also on AO website. These show the total hours covering particular frequency range, and the fraction of that time occupied by RFI.

RFI Mitigation:

An example was shown of the OH spectrum of an OH/IR star obtained with 3-level sampling using the present autocorrelator showing lots of ringing. The effects of RFI are helped significantly by 9 level sampling.

Same data stream processing differently can also mitigate for RFI, e.g. median filtering. Algorithm develoment will be important It may be ossible to use algorithms to extract cleaned spectra. Eg. Simulated data, recorded in voltage (before detection), calculate variance, inject signal in gaussian stream, add non-gaussian noise (poissonian) to simulate RFI. Use algorithm to do weighting, and recovering signal nicely.

Arecibo anti-RFI activity: importance of L-band to ALFA:

Several issues/questions are relevant to E-ALFA:


Karen O. What is lowest frequency of the radar pair at Puntas Salinas?
Tapasi G. 1223 MHz.
Wim vD. There are no unique RFI mitigation algorithms; it is telescope dependent. One approach certainly is to employ radar blankers. But also, we need simple RFI removal software. GPS, for example, is not on all the time, but you need software to excise it from scans. You also need to consider: "How short should your sampling rate be, so that you can excise RFI?". The latter is critical for correlator concept.

In order to assess how far out in z (low in freq) we can go, we have to know how clean is the spectrum. This also relates to the issue of maximum BW: 100 or 200 MHz bandwidth? From my experience, the spectrum seems to be pretty clean out to 1225 MHz, and so there is a case for 200 MHz bandwidth.
Phil P. You also need to be sure not to saturate the analog signal. For example, Aerostat 126 can blow out a digitizer. Sometimes you may want to use multiple pieces of narrow bandwidth each so you don't saturate the whole band.
Riccardo Wide field surveys are not going to have extremely long integration times, so they probably will not focus on detection of signals over wide bandwidth. We have to look at the tradeoffs between bandwidth coverage versus potential problems.
Wim The point is that it is not entirely clear.
Wolfram F. We also need software to excise to cross-correlate beams to remove RFI.
Lister Clearly cross-correlation can help.
Riccardo There are a bunch of questions that we need to investigate. Do you clean first in time freq domain? (Yes). Is it critical to make multiple passes?
Lister Median filtering for given pixel on sky is the only nonparametric method used on HIPASS. It is quite robust. It doesn't get rid of extra rms around GPS L3, though it does get rid of the L3 signal itself.
Noah B. Can you coordinate movement of frequency of radars?
Riccardo Yes, at least for the San Juan airport radars. For example, Wolfram did it all the time in order to make his Hercules supercluster observations.

next up previous
Next: The GALFA Surveys Up: Minutes of the 1st Previous: The WAPP

This page created and maintained by Karen Masters, Kristine Spekkens and Martha Haynes.

Last modified: Thu Apr 17 18:07:14 EDT 2003