RACON interference with bird radar

Mitigation

We can try to reduce interference to/with RACONs by these methods:

• move bird radar farther away from the RACON
• set bird radar to "sector blank" in the direction of the RACON; i.e. don't transmit pulses in a sector centred on the RACON
• put bird radar well above or below the RACON, so the radar's beam is never directly pointing at the RACON
• rotate the polarization of the bird radar from horizontal to vertical.  Most X-band RACONs whose details we're aware of have horizontally polarized antennas, matching that of most X-band ship radars.  For a typical bird radar, this rotation requires re-orienting the rectangular waveguide feed so that the long edge (in cross-section) is vertical, rather horizontal.  We don't know how much bird radar cross-sections will differ between the two polarization modes, but it shouldn't significantly affect our ability to detect birds since a vertically operated t-bar antenna is (briefly and repeatedly) vertically polarized when it is aligned up-and-down, and we haven't heard of or noticed any corresponding "disappearance" of birds from the radar at those times.

Lab test of polarization rotation

We fed a Furuno FR-1964C-BB with 62cm parabolic dish pulses from a Wavetek 907A signal generator via a coaxially-fed horn 2 metres away from the dish at closest approach.  The radar's video signal was digitized using a modified USRP-1 from Ettus research, and data were gathered using radR.  Several combinations of signal power and horn antenna orientation (horizontal or vertical) were used, and for each, at least 50 sweeps were monitored, and for each, the 99-th percentile of the brightest fixed region of the pulse x sample matrix was calculated.  A linear model was used to estimate the effects of signal power and horn antenna polarization on the digitized values.  The dataset is attached below.

The results look like this:

> summary(lm(s99~power+orient, dat))
...
Residuals:
    Min      1Q  Median      3Q     Max
-68.173 -10.136  -3.203  19.432 178.432

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept) 3513.5768     4.1520   846.2   <2e-16 ***
power         32.5135     0.1417   229.5   <2e-16 ***
orientv     -551.1719     3.1365  -175.7   <2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 37.37 on 905 degrees of freedom
Multiple R-Squared: 0.9841,    Adjusted R-squared: 0.9841
F-statistic: 2.804e+04 on 2 and 905 DF,  p-value: < 2.2e-16

> -551.1719 / 32.5135
[1] -16.95209

The division of the power coefficient by the orientation (orientv) coefficient shows that rotating the polarization of the microwave signal from horizontal to vertical reduces the effective signal strength at the radar receiver by approximately 17dB, i.e. a factor of 50, or the same effect as increasing the distance between RACON and radar by a factor of 7 (both interference effects are due to one-way signal trips, so power drops with the square of distance).  This is seen as a 17dB horizontal separation between (colour) groups in the corresponding scatter plot:

The graph is a bit confusing:  the x-axis represents transmitted power, while the y-axis represents received power.  Received power is approximately 500 sample units higher when the horn transmitter is horizontally polarized (i.e. matches the radar) than when it is vertically polarized (i.e. is mis-matched to the radar).  A difference of 500 sample units along the y-axis corresponds to a difference of approximately 17 dB along the x-axis, according to the linear model (i.e. if you move along the regression lines).

Partial List of RACON sites around Nova Scotia

We should try the polarization rotation in a field situation with a live RACON.  The following list is from the Canadian Coast Guard Notices to Mariners available here  (list of Lights, Buoys, and Fog Signals):

Live test at Chebucto Head Lighthouse+Racon site (26 October 2011)

We inlined a twisted waveguide onto the Furuno 1954C with tilter unit and made an x-braced offset to accomodate it.  The feed waveguide and horn were rotated 90 degrees to match the new waveguide as shown here:

The field test took place at 11:40 ADT under these weather conditions at Halifax, courtesy of Environment Canada:
 

Wind precluded setting up the radar on a tripod, as originally planned, so it was placed on the ground, braced against the wind by a concrete bulkhead.  The radar was a Furuno FR1954-CBB 12.5kW operating in short pulse (80ns) mode at PRF=2100 Hz. We used a modified Ettus Research USRP-1 to digitize the radar signal, which was recorded with radR.  These photos illustrate the set-up:
       

The following sweep-by-sweep movie shows our radar triggering the racon during the entire 360 degree sweep.  It appears the racon has a latency period or some other protection so that at least one sweep failed to trigger it.  We only tested in vertical polarization mode (i.e. with the 90-degree twisted waveguide installed) because the results with this mitigation were clearly negative, so that no control trial in horizontal polarization mode seemed necessary.

Hit ESC to stop the animation.  Movie playback is at 5x speed - timestamps are in the plot titlebar.  Note the radial DASH DASH DOT DOT pattern (Morse code 'z') which identifies the Chebucto Head beacon.