Signal Processing Challenges in Electronic Surveillance (ES)

pancakesbootΤεχνίτη Νοημοσύνη και Ρομποτική

24 Νοε 2013 (πριν από 3 χρόνια και 8 μήνες)

112 εμφανίσεις

UNCLASSIFIED

UNCLASSIFIED

Signal Processing Challenges
in Electronic Surveillance (ES)

20 March 2009


David Nethercott


EW & Nav group

DSTL

© Dstl 2009


Dstl is part of the
Ministry of Defence


UNCLASSIFIED

UNCLASSIFIED

20 March 2009

Signal Processing Challenges in
Electronic Surveillance


Electronic Intelligence (ELINT)


traditionally the gathering of
information by intercept of RADAR
transmissions


RADAR systems usually pulsed and
high power


Intercept of RADAR signals is often
at high SNR


Air surveillance RADAR

© Crown Copyright

© Dstl 2009


Dstl is part of the
Ministry of Defence


UNCLASSIFIED

UNCLASSIFIED

20 March 2009

Signal Processing Challenges in
Electronic Surveillance


Problems:


RADAR spectrum is very
congested due to proliferation of
systems (military and civil)


Multiple overlapping transmissions
from multiple sources


Need to identify and isolate these
different transmissions


Duty cycle of observations can be
limited (
~
10%)

Man
-
portable pulse Doppler RADAR

© Crown Copyright

© Dstl 2009


Dstl is part of the
Ministry of Defence


UNCLASSIFIED

UNCLASSIFIED

20 March 2009

Signal Processing Challenges in
Electronic Surveillance


Signal description


SNR is not a problem


Overlapping signals in time
and frequency


Often pulsed, but changing
from pulse to pulse


Complex modulation
schemes


eg comms
channels

time
frequency


100
200
300
400
500
300
350
400
450
500
550
600
650
700
750
800
40
45
50
55
60
65
70
75
80
85
90
© Crown Copyright

© Dstl 2009


Dstl is part of the
Ministry of Defence


UNCLASSIFIED

UNCLASSIFIED

20 March 2009

Signal Processing Challenges in
Electronic Surveillance


Challenges:


FFT and linear filtering is good for separation of narrowband
signals with no frequency overlap


but how to tackle broadband
overlapping signals?


Signals evolve over time (from pulse to pulse)


Complex modulation schemes, scope not limited to simple
RADAR transmissions


Real time and high data rates


Low duty cycle of observations



© Dstl 2009


Dstl is part of the
Ministry of Defence


UNCLASSIFIED

UNCLASSIFIED

20 March 2009

Signal Processing Challenges in
Electronic Surveillance


Past approach has used Hough Transform


Current approaches use FFTs and tracking algorithms to follow
energy in time/frequency space.


Kalman tracker of thresholded energy values in TFR


Rule based parameterisation of TF features


This approach works well, but:


computationally intensive


prior assumptions about the signals must be made, eg: sweep linearity



What signal decomposition techniques could be applied?


© Dstl 2009


Dstl is part of the
Ministry of Defence


UNCLASSIFIED

UNCLASSIFIED

20 March 2009

Points for discussion


Multiple signal decomposition methods:


PCA/SVD


Statistical based, computationally intensive


EMD


computationally intensive, good signal decomposition, yields
temporal nature of signals, eg: amplitude/frequency


Other signal characterisation methods:


Instantaneous amplitude/frequency through Hilbert transform


works well for single signals, but multiple signals?


© Dstl 2009


Dstl is part of the
Ministry of Defence


UNCLASSIFIED

UNCLASSIFIED

20 March 2009

Points for discussion


Short duty cycles of observation:


What information can be obtained with limited observations?


For complex modulation schemes, eg: digital comms, how much
of the signal is required to obtain:


Modulation type, eg: QPSK, BPSK, OFDM etc.


Modulation characteristics, eg: baud rate


Localisation or direction finding