5GHz RLAN Interference on Active Meteorological Radars

1
5GHz RLAN Interference on Active Meteorological
Radars

• Presenter André L. Brandão
• CRC Ottawa
• andre.brandao_at_crc.ca

2

• Subject
• Interference of
• RLAN Meteorological Radars
• Although there is mutual interference, in this
  work we discuss solely the effects of
  RLAN-to-radar.

3
(No Transcript)
4
(No Transcript)
5
(No Transcript)
6
(No Transcript)
7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
EXPERIMENTAL

• The team
• (CRC) André L. Brandão, John Sydor, Wayne Brett
• (EC) John Scott, Paul Joe, Derek Hung
• Summer of 2004 CRC and EC conducted field
  experimentation that examined the effects of WAS
  into weather radars. Meteorological observations
  enable forecast and warning services that are
  essential for public safety and RLAN interference
  could jeopardize the normal operation of weather
  radar stations.

11

• A - SCENARIO
• Conventional radars map location and intensity of
  precipitation. The Doppler weather radar adds
  functionality (motion and structure within storm
  formations).
• The Canadian weather network 31 Doppler radars
  C-Band (co-axial magnetron transmitter tubes,
  tuned at 5625 MHz with a tuning range from 5450
  MHz to 5825 MHz.
• Parabolic dishes (beamwidth of 0.65deg). Operate
  in pulse mode with pulse durations of 0.8, 2 and
  10microsec. Pulse freq. from 1.2 kHz to 50 Hz
  (maximum sustained duty cycle for the magnetron
  is 0.1).
• Transmitter peak power of 250 kW. Radars operate
  in continuous azimuth scanning mode. Complete
  volume scan every 5 to 10 minutes (depending on
  scan type).
• The Franktown CWSR98 weather station has
  effective detection range up to 240 kilometres.

12
(No Transcript)
13
(No Transcript)
14

• B METHODOLOGY
• The experiment consisted of a series of tests
• (a) collection of radar products (gif ascope)
  when RLAN signals were not present (i.e. RLAN
  aggregate power at levels ?110dBm or below, as
  measured at the CWSR98 antenna output port)
• (b) collecting data when RLAN packets were input
  to the radar with various levels of power (i.e.
  above ?110dBm), and a variety of modulations,
  pulse rates and carrier frequencies.
• Data gathering for (a) interlaced with (b). RLAN
  ON/OFF switched lt 30s.
• Interference power levels of -110dBm used as
  criteria for RLAN OFF (limit of the CWSR98
  sensitivity)
• Other radar systems can benefit from our results
  but should use their own radar sensitivity
  settings and apply our findings accordingly.

15
(No Transcript)
16
C - RESULTS AND DISCUSSION

• 1) INJECTION TESTS
• RLAN packets injected into RF front-end of radar
  (5.61GHz).
• variants
• inter arrival time between packets (1500 bytes
  per packet payload)
• modulation (BPSK, QPSK up to 64QAM) signal
  power and carrier frequency.
• Ascope data (as computed by the radar signal
  processor) collected for each test.
• Total reflectivity parameter extracted. Based on
  the transmit radar pulse characteristics and
  corresponding filters used, amongst other
  factors, we could compute the radar noise power
  and theoretical minimum signal-to-noise ratio
  (SNR), assuming a 290K system noise temperature.

17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20

• RLAN 57dBm and RLAN 47dBm (over BW of 18MHz)
  are reflectivity values computed by radar with
  interference present and measured at the 30MHz IF
  input to the A/D converter of the radar signal
  processing module.
• With RLAN 67dBm the effect of the interference
  into the radar processing is negligible. Radar
  input filter is 1.13MHz
• (1)
• PRLAN is the effective RLAN interference power
  measured at the radar processor
• Comparing the result with the theoretical value
  from table we verify that the nominal noise power
  of 80.95 dBm is very close to PRLAN found
  experimentally.
• Preliminary conclusion is RLAN signals cause
  detectable radar signal degradation when the
  aggregate interference power achieves a level
  close to the nominal SNR of the radar.

21

• A direct consequence of the result RLAN
  interference acts as uncorrelated additive noise.
  
• This result may now appear obvious, but remember
  that prior to field trials such a conclusion was
  far from certain.
• Generally, signal vectors corrupted by
  uncorrelated additive noise have their magnitude
  and phase distorted at the same time. Thus, phase
  degradation of return radar pulses should
  manifest itself similarly at the same time
  reflectivity errors become apparent.
• Phase and frequency information are used to track
  the speed and direction of rain cells.
• Therefore this result is important for finding a
  threshold value for phase and frequency (Doppler
  shift computation) disturbances.

22
The effect of RLAN duty cycle

• Duty cycle of the RLAN interference was varied by
  increasing the interarrival time of fixed length
  packets (750 ms total packet duration).
• The CWSR98 set in operational mode of 0.8microsec
  pulse duration and 1.13 MHz IF bandwidth.
• The average packet power was set at -20 dBm
  (measured at the 30 MHz IF), which produced an
  18.57 dBz mean reflectivity reading by the radar
  signal processor (we used this as our reference).
  
• The radar was not transmitting. Only background
  noise and interference were responsible for the
  mean reflectivity reading.
• The duty cycle of the RLAN was then varied, and
  with each variation, the mean reflectivity was
  recorded.

23
(No Transcript)
24
(No Transcript)
25
2) RADIATED INTERFERENCE (OUTDOOR TESTS)

• Outdoor tests the generator was connected to a
  10 watt classA linear power amplifier (then via
  a calibrated cable, 11dB loss to a 12 dBic
  circularly polarized antenna, mounted on an
  adjustable hydraulic mast)
• Using this technique a signal of up to 38 dBm
  EIRP (per polarization) was capable of being
  generated.
• By regulation, a RLAN device is not allowed to
  radiate more than 30 dBm. The tests were
  conducted in Carleton Place, Ontario, Canada,
  located 10.6 km from radar.
• The RLAN antenna raised to heights 15 m and 7 m
  respectively.
• For the first height the path loss exponent (PLE)
  2.8
• For the 7 m height the PLE was close to 3.2.

26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29

• One contribution of this work is the confirmation
  that RLAN signals indeed appear as straight lines
  across the weather map.
• The results can be used as part of the
  identification process for the signature of a
  RLAN signal mixed within the radar processing
  unit.
• Nevertheless, this feature alone is not
  sufficient for determining if a RLAN signal is
  present or not in the system.
• Artefacts such as blockage generate streaks over
  the image also. Blockage occurs when trees, roof
  tops, etc., are situated close to the antenna and
  partially obstruct the radar pulse in that
  direction.
• When this happens, the echo powers within the
  obstructed area return with a mean power level
  below the average power for adjacent areas.
• RLAN streak is caused when the mean power level
  is above the mean value of a corresponding area
  alongside the streak.
• This is only if RLAN are incoherent with radar
  pulses and modelled as additive noise.

30
(No Transcript)
31
CONCLUSIONS

• Our experiments have quantified sensitivity
  threshold of a Doppler weather radar system to
  RLAN (IEEE 802.11a) interference. Based on CWSR98
  radar, we could conclude that the mechanism of
  degradation seems to be one which is related
  purely to the mean amount of interference power
  deposited by the RLAN on the radar signal. To the
  radar, RLAN will manifest itself mainly as
  additive uncorrelated noise independent of
  modulation, power, baud rate and packet
  interarrival time.
• For the CWSR98 weather radar we found an
  experimental aggregate RLAN power of 79 dBm as
  the threshold value for reflectivity degradation.
  That matches closely with the amount of noise
  power used in the computation of the nominal
  minimum signal-to-noise ratio for that system.
• The result is also important for estimating
  threshold values for phase and frequency
  disturbances (Doppler shift computation).
  Finally, other radar operators can rely on their
  computed nominal SNR sensitivity thresholds for
  estimating the maximum aggregate RLAN power that
  their systems can tolerate. Such estimates can be
  used as a basis in determining whether DFS will
  be sufficient to prevent interference by
  individual and aggregate RLAN power found within
  the beam of the radar.