Processing the data:

• There were 71 scans (92069 1 second spectra).
• Only  data from alfa beam 0 was used.
  
• For each scan +/- 2 MHz about 1381.05 MHz (164 channels, 24 Khz resolution) was extracted and then used for the following processing.
• PolA and polB were averaged together.
  
• For each 1 second average spectrum:
• .75 MHz on  each edge of the 4MHz  band was averaged and then used to normalize the spectra to Tsys. This got rid of celestial sources.
• Total power for each 1 second was then computed using +/- .3 MHz around 1381.05
• 1. was subtracted from each total pwr sample.
  
• Any total power points above .02 Tsys were taken to be rfi from gps L3

The plots show the processing steps (.pdf) (2.9 Mbytes)

• Top frame: Frequency limits used to compute total power about L3 band.
• 2000 1 sec spectra were over plotted showing the gps L3 rfi
• The dashed red lines are the frequency limits used to compute the total power around the L3 line.
• Bottom frame: total power samples for the month.
  
• each + is the +/- .3 MHz total power computed about 1381.05. The units are Tsys (i've removed 1.)
• there are 92000 samples (the plot clips at .05 Tsys)
• The red line is the threshold used to define gps L3 rfi.
• There were 8346 points above the line (not all shown)
• The threshold ignores a number of L3 points (eg around 55,000)
• 8346/92000 is about 9% usage.

Images and histograms of the month:

   There were 78 scans for the month. 76 of these were 1200 seconds long. For each 1200 secs scan an image was made  covering 1200 secs,  +/- 2 MHz about the gps L3 frequency

A mosaic was made of 76 1200  scans (.png)

• 76 1200   scans were available (two shorter scans were ignored). This is 91200 1 sec spectra.
  
• Each frame is 1200 seconds (vertical) by 4 MHz (horizontal)
  
• The bright lines in the central portion of a frame comes from gps L3.
• The scan number (on each title line) has Ydddnnnnn. Where Y is 6 for 2016, ddd is dayno of the year (ast), and nnnnn is a number that increments within a day. This shows which scans were taken on the same day.
• 51 of the 76  1200 second  scans had gps interference.
• Of the 91200 1 sec spectra, 8346 had L3 interference  9% of the time. This used the  .02 threshold above... this underestimated the rfi by maybe 100.

Bar chart  of the rfi by scan (.png)

• The vertical scale is the samples within in a scan
• the horizontal axis is the day number of jul16 when the scan was taken
• The colors show good and bad data
• red: these samples had gps L3 interference
• green: samples with no gps L3 interference.
• I included the 2 shorter scans in the bar chart.
  

SUMMARY:

• 76 1200 second scans were taken with the ALFA receiver by A2048 during july 2016.
• Beam 0 data was used to look for gps L3 interference.
• 91200 total 1 second spectra were used.
• 8346 1 sec spectra had gps L3 interference (using the .02 Tsys threshold on the total power)
• L3 interference was seen 9% of the time.
  
• 51 of the 76 scans had L3 interference 67%.
• There were no scheduled gps L3 tests during the month of jul 2016.
• there are a number of 200 sec blocks of rfi..
  
• i should probably go through and see if there is a particular satellite that is causing more than average rfi.
• (but that will take some more work...)
  

page up
home ~phil