30-06-2012, 05:03 PM
FM radio based bistatic radar
[attachment=26358]
Introduction
In this paper we describe an experimental radar system
developed over the past eighteen months that detects and
tracks aircraft by receiving and processing echoes from a
single non-cooperative frequency modulated (FM) commercial
radio station.
Of all the transmitters of opportunity available in the
environment, broadcast transmitters represent some of the
most attractive for surveillance purposes, owing to their
high powers and excellent coverage.
At first glance, analogue television transmitters seem the
obvious choice of illuminator, as they have very high
equivalent radiated powers. However, despite the instant
appeal of their pulse-like waveform structure, it is quickly
found that the waveform is far from suited for radar usage
when used in a conventional radar matched filtering
approach [1]. Howland showed, however, that it is possible
to exploit the Doppler and bearing information in echoes of
the television video carrier signal to track aircraft at ranges
of up to 260 km from the receiver and 150 km from the
transmitter [2].
Wideband processing
By contrast, ‘wideband processing’ is defined in passive
radar systems as the use of a receiver bandwidth that is
comparable to the bandwidth of the waveform being
exploited. For example, a typical FM radio broadcast
occupies a bandwidth of about 100 kHz. Radar range
resolution is approximately equal to c=2B and hence we
see that the FM broadcast offers a potential range resolution
of up to 1500 m. It is clear that the signal offers useful target
ranging information. Although we focus on the use of FM
radio signals in this paper, it should be noted that the
approach described here is generic and applicable to any
transmission of opportunity with a reasonable ambiguity
function, such as cell-phone transmissions or digital radio or
television waveforms.
Expected system performance
In our system we are exploiting a single, vertically-polarised
FM radio transmitter located at Lopik, some 50 km behind
the receiver. The transmitter has a mean ERP of 50kW and
frequency of 96.8 MHz. The transmitter is located on a
375m mast and provides excellent long-range low-level
illumination.
The receiver is located on the roof of the NATO C3
Agency in The Hague, and is approximately 20m above
ground level. The Agency is on the edge of some sand dunes
which lead to the North Sea approximately 2 km away.
The receiver comprises two vertically-polarised half-wave
dipoles over a wire-mesh backplane 1.5 wavelengths by 1.5
wavelengths in size. The receiver antenna is steered so as to
try to place the transmitter in a null in the antenna pattern to
reduce the unwanted direct signal.
System overview
The system described in this paper was built on a low budget
and is one of the simplest architectures that can be used to
explore this technology. A block diagram of the system
hardware is shown in Fig. 3 and of the processing algorithm
in Fig. 4.
Reading from left to right, the signal is collected by a
digital receiver system comprising of at least three channels.
This allows for one reference channel and two surveillance
channels for direction finding. An adaptive filter is applied
to the two surveillance channels to reject the unwanted
transmitter signal and then the digital data from the three
channels are fed to the cross-correlator that outputs two
amplitude–range–Doppler (ARD) surfaces.
Adaptive removal of the direct signal and surface clutter
Although the cross-correlation processing between the
reference and surveillance channels causes any unwanted
reference signal in the surveillance channel to be confined to
the zero-Doppler and zero-range bin, the range and Doppler
sidelobes of this autocorrelation function remain significant.
At best, with a 1 second integration time and 50 kHz
effective bandwidth, these will be 47 dB below the main
autocorrelation peak. However, given that the direct signal
may be 80–90 dB greater than the echoes themselves, this
means that the sidelobes remain some 30–40 dB higher than
the echoes we are seeking. This is compounded by strong
surface clutter returns from the sea surface to a bistatic
range of around 50 km.
An efficient implementation
The major drawback of the approach presented above is the
excessive processing load owing to calculations of the Fast
Fourier Transforms for long input signals. We resolve this
issue by applying a decimation technique that allows us to
discard data at Doppler frequencies we know targets do not
exist, before calculating the Fourier transform. This
modified integration algorithm utilises some extra processing
steps to decimate the signal but greatly reduces the
overall computation complexity with almost no loss in
signal processing gain.
[attachment=26358]
Introduction
In this paper we describe an experimental radar system
developed over the past eighteen months that detects and
tracks aircraft by receiving and processing echoes from a
single non-cooperative frequency modulated (FM) commercial
radio station.
Of all the transmitters of opportunity available in the
environment, broadcast transmitters represent some of the
most attractive for surveillance purposes, owing to their
high powers and excellent coverage.
At first glance, analogue television transmitters seem the
obvious choice of illuminator, as they have very high
equivalent radiated powers. However, despite the instant
appeal of their pulse-like waveform structure, it is quickly
found that the waveform is far from suited for radar usage
when used in a conventional radar matched filtering
approach [1]. Howland showed, however, that it is possible
to exploit the Doppler and bearing information in echoes of
the television video carrier signal to track aircraft at ranges
of up to 260 km from the receiver and 150 km from the
transmitter [2].
Wideband processing
By contrast, ‘wideband processing’ is defined in passive
radar systems as the use of a receiver bandwidth that is
comparable to the bandwidth of the waveform being
exploited. For example, a typical FM radio broadcast
occupies a bandwidth of about 100 kHz. Radar range
resolution is approximately equal to c=2B and hence we
see that the FM broadcast offers a potential range resolution
of up to 1500 m. It is clear that the signal offers useful target
ranging information. Although we focus on the use of FM
radio signals in this paper, it should be noted that the
approach described here is generic and applicable to any
transmission of opportunity with a reasonable ambiguity
function, such as cell-phone transmissions or digital radio or
television waveforms.
Expected system performance
In our system we are exploiting a single, vertically-polarised
FM radio transmitter located at Lopik, some 50 km behind
the receiver. The transmitter has a mean ERP of 50kW and
frequency of 96.8 MHz. The transmitter is located on a
375m mast and provides excellent long-range low-level
illumination.
The receiver is located on the roof of the NATO C3
Agency in The Hague, and is approximately 20m above
ground level. The Agency is on the edge of some sand dunes
which lead to the North Sea approximately 2 km away.
The receiver comprises two vertically-polarised half-wave
dipoles over a wire-mesh backplane 1.5 wavelengths by 1.5
wavelengths in size. The receiver antenna is steered so as to
try to place the transmitter in a null in the antenna pattern to
reduce the unwanted direct signal.
System overview
The system described in this paper was built on a low budget
and is one of the simplest architectures that can be used to
explore this technology. A block diagram of the system
hardware is shown in Fig. 3 and of the processing algorithm
in Fig. 4.
Reading from left to right, the signal is collected by a
digital receiver system comprising of at least three channels.
This allows for one reference channel and two surveillance
channels for direction finding. An adaptive filter is applied
to the two surveillance channels to reject the unwanted
transmitter signal and then the digital data from the three
channels are fed to the cross-correlator that outputs two
amplitude–range–Doppler (ARD) surfaces.
Adaptive removal of the direct signal and surface clutter
Although the cross-correlation processing between the
reference and surveillance channels causes any unwanted
reference signal in the surveillance channel to be confined to
the zero-Doppler and zero-range bin, the range and Doppler
sidelobes of this autocorrelation function remain significant.
At best, with a 1 second integration time and 50 kHz
effective bandwidth, these will be 47 dB below the main
autocorrelation peak. However, given that the direct signal
may be 80–90 dB greater than the echoes themselves, this
means that the sidelobes remain some 30–40 dB higher than
the echoes we are seeking. This is compounded by strong
surface clutter returns from the sea surface to a bistatic
range of around 50 km.
An efficient implementation
The major drawback of the approach presented above is the
excessive processing load owing to calculations of the Fast
Fourier Transforms for long input signals. We resolve this
issue by applying a decimation technique that allows us to
discard data at Doppler frequencies we know targets do not
exist, before calculating the Fourier transform. This
modified integration algorithm utilises some extra processing
steps to decimate the signal but greatly reduces the
overall computation complexity with almost no loss in
signal processing gain.