sensing and communications using ultrawideband random noise waveforms professor ram m. narayanan...

Sensing and Communications Sensing and Communications Using Ultrawideband Random Using Ultrawideband Random

Noise WaveformsNoise Waveforms

Professor Ram M. NarayananProfessor Ram M. NarayananDepartment of Electrical EngineeringDepartment of Electrical Engineering

The Pennsylvania State UniversityThe Pennsylvania State UniversityUniversity Park, PA 16802, USAUniversity Park, PA 16802, USA

Tel: (814) 863-2602Tel: (814) 863-2602Email: [email protected]: [email protected]

2005 AFOSR Program Review for Sensing, Imaging and 2005 AFOSR Program Review for Sensing, Imaging and Object RecognitionObject Recognition , Raleigh, NC, May 26, 2005, Raleigh, NC, May 26, 2005

May 26, 2005May 26, 2005 2005 AFOSR S&S Program Review2005 AFOSR S&S Program Review 22

OutlineOutline

• Introduction• Why use noise waveforms• Noise waveform modeling• Heterodyne correlation approach• Polarimetric radar applications• Radar imaging applications• Covert communications applications• MIMO network concept• Conclusions


IntroductionIntroduction

• Military operations require low probability of intercept (LPI), low probability of exploitation (LPE), low probability of detection (LPD), and anti-jam characteristics

• Traditional radar and communications systems use conventional deterministic waveforms

• Deterministic waveforms (such as impulse/short-pulse and linear/stepped frequency modulated) do not possess above desirable features


Why use noise waveforms?Why use noise waveforms?• Noise waveforms are inexpensive to generate both in

analog and digital formats• Noise waveforms have featureless LPI/LPD characteristics

and are therefore covert• Noise waveforms are inherently anti-jam and interference

resistant• Noise sources are easily obtained using current microwave

and RF circuit technology• Noise waveform spectral characteristics can be adaptively

shaped to suit the dynamic environment• Noise waveforms are spectrally very efficient and can share

spectral bands without mutual interference• Noise waveforms exhibit excellent waveform diversity

characteristics


Waveform comparisonWaveform comparison

0 500 1000 1500-0.5

0

0.5

1Impulse waveform

Am

plitu

de

0 500 1000 1500-1

-0.5

0

0.5

1LFM waveform

Am

plitu

de

0 500 1000 1500

-2

0

2

Random noise waveform

Time

Am

plitu

de


Simple noise radar architecture Simple noise radar architecture using homodyne correlatorusing homodyne correlator


Phase coherence injection Phase coherence injection

• Homodyne correlation noise radar downconverts directly to DC and hence loses important phase information of returned signal

• There is a way to inject phase coherence in noise radar using time-delayed and frequency-offset transmit replica

• Heterodyne correlation noise radar downconverts to offset frequency and preserves phase information of returned signal


Noise waveform - stochastic modelNoise waveform - stochastic model

• Thermal noise is stochastic and can therefore only be described by its statistics

• Noise signal x(t) can described as follows: PDF px(X) ► Zero-mean Gaussian

Autocorrelation Rxx(τ) ►Impulse at τ = 0

PSD Sxx(f) ►White, assumed uniform and bandlimited

• Above representation does not permit time-frequency equivalence for tracing the signal through the system


Noise waveform – time-frequency Noise waveform – time-frequency modelmodel

})](cos{[)()( 0 tttatx

where• a(t) is Rayleigh distributed amplitude that describes

amplitude fluctuations• δω(t) is uniformly distributed frequency that describes

frequency fluctuations [-Δω ≤ δω ≤ +Δω]• average power = ½‹a2(t)›/R0, assuming a(t) and δω(t)

are uncorrelated• center frequency = ω0/2π = f0

• bandwidth = 2Δω/2π = B


Bandwidth descriptorsBandwidth descriptors

•Narrowband ► B/f0 ≤ 10%

•Ultrawideband (UWB) ► B/f0 ≥ 25%

Although time-frequency representation is inherently narrowband, we extend it to the UWB case owing to its simplicity and ease of signal

analysis


Alternate time-frequency Alternate time-frequency representationrepresentation

)2sin()()2cos()()( 00 tftstftsts QI

)](2cos[)()( 0 ttftats

)()()( 22 tststa QI

where

)(

)(tan)( 1

ts

tst

I

Q

where sI(t) and sQ(t) are zero-mean Gaussian processes and

f0 is the center frequency

This can be recast as

Rayleigh distributed

Uniformly distributed


Homodyne correlation noise radarHomodyne correlation noise radar

Noise Source

PowerDivider

TimeDelay

Output

MixerAntenna

Antenna


Heterodyne correlator noise radarHeterodyne correlator noise radar

NoiseSource

PowerDivider

TimeDelay

Output

Mixer

Antenna

Antenna

LSB Upconverter

OffsetFrequency

Source


Heterodyne correlation noise radar Heterodyne correlation noise radar signal analysissignal analysis

• Transmit waveform ►

• Received waveform ►

• Time-delayed transmit replica ►

• Time-delayed and frequency-offset transmit

replica ►

• Low-pass filtered correlator output when both

delays match (zero otherwise) ►

where Γ and Θ are magnitude and phase of target reflectivity, t0 and td are target and internal

delays, and ω′ is the offset frequency

}]cos{[)()( 0 ttatvt

)}](cos{[()()( 0 ddd ttttatv

}))](cos{[()()( 000 ttttatvr

)}](cos{[()()( 0 ddd ttttatv

]cos[)()( 2 avo


Coherent reflectivity extractionCoherent reflectivity extraction

• Output of correlator is ALWAYSALWAYS at offset frequency!!• UWB transmit waveform collapses to a single frequency!• We can shrink detection bandwidth at correlator output

to enhance SNR

• Power in correlator output is proportional to Γ2

• I/Q detector in receiver can measure Θ• Doppler, if any, will modulate correlator output and can

be extracted from the I/Q detector• Offset frequency usually lies between 10-15% of center

frequency of transmission


What can coherency give us?What can coherency give us?

• Polarimetry• Interferometry• Doppler estimation• SAR imaging• ISAR imaging• Monopulse tracking• Clutter rejectionALL USING INCOHERENT NOISE RADAR!!!


Difficulty of stochastic representationDifficulty of stochastic representationGenerate random signal s(t)

Calculate its Fourier Transform S(ω)

Generate the offset-frequency Fourier

Transform S(ω-ω′)

Generate the reflected signal Fourier Transform

Γexp(-jΘ)S(ω)

Multiply above signals and perform low-pass filtering

Compute its Inverse Fourier Transform


Dual-channel polarimetric noise Dual-channel polarimetric noise radar architectureradar architecture

OSC1PD1

OSC2PD2

DL1

DL2

MXR1

PD4

AMP1

PD6

MXR2

MXR3

ANT1

ANT2

FL1 FL2

Co-polarizedI/Q

Co-polarizedAmplitude

Cross-polarizedAmplitude

AMP3

PD3

AMP2

AMP4

AMP5

AMP6

AMP7

PD5

Cross-polarizedI/Q

IQD1

IQD2


Time domain Frequency domain

Bandlimited noise waveformBandlimited noise waveform


Measured point spread functions Measured point spread functions of Channel 1 and Channel 2of Channel 1 and Channel 2


Approximate resolutionsApproximate resolutions

• Range resolution

where c is speed of light and B is the transmit bandwidth

• Doppler resolution

where Tint is the integration time

BcR 2

int

1Tfd


B = 1 GHz

Tint = 50 (L), 10 (R) ms

Average ambiguity functionsAverage ambiguity functions

B = 100 MHz

Tint = 50 (L), 10 (R) ms


Application examplesApplication examples

• Ground penetration imaging• Arc-SAR imaging• Polarimetric ISAR imaging• Foliage penetration (FOPEN) SAR imaging• Anti-jamming imaging performance


Detection of multiple objects: Two metallic plates, 17.8 cm and 43.2 cm depth

Ground penetration imagingGround penetration imaging


Detection of non-metallic object: Distilled water in 1 gallon plastic container, depth 7.6 cm



Detection of polarization-sensitive object: Metallic pipe, parallel to transmit polarization and parallel to scan direction



Ground penetrating imagingGround penetrating imagingDetection of polarization-sensitive object: Metallic pipe, parallel to transmit

polarization and perpendicular to scan direction


Arc-SAR imagingArc-SAR imagingSAR image of two corner reflectors using a 1-2 GHz random noise radar


RGB color composite image of mock airplane (Red=HH, Green=HV/VH, Blue=VV)

Geometry of mock airplane

Polarimetric ISAR imagingPolarimetric ISAR imaging


Images of two trihedral reflectors under foliage coverage, HH polarization

Trihedral-1Trihedral-2

Trihedral-1

Trihedral-2

Tree-1

Tree-4

Tree-2

Tree-3

Tree-1

Tree-2 Tree-3Tree-4

Target scenario FOPEN SAR image SVA enhanced image

FOPEN SAR imagingFOPEN SAR imaging


Simulated ISAR images of a MIG-25 airplane: no jamming (top), LFM radar image with SJR = -10 dB (top right), and noise radar image with SJR = -10 dB (right)

Anti-jamming imaging performanceAnti-jamming imaging performance


Covert communications conceptual Covert communications conceptual architecturearchitecture

NoiseSource

PowerDivider

Modulator

MessageSignal

LSB Mixer

Channel 1

Channel 2

De-Modulator

MixerOutput

Transmitter ReceiverChannel 1 is the “key”


Diversity implementationsDiversity implementations

• Polarization diversity: Channels 1 and 2 transmitted over orthogonal polariztions

• Band stacking (Frequency diversity): Channels 1 and 2 are made to occupy contiguous spectral bands

• Delay diversity (Time diversity): Channel 2 delayed and transmitted after Channel 1


Diversity implementation featuresDiversity implementation features


Polarization diversityPolarization diversityTransmit waveformsTransmit waveforms

• Noise source output ►

• Horizontally transmitted waveform ► ► NoiseNoise

• Modulator output ►

• LSB mixer output ►

• Vertically transmitted waveform ►

► Noise-likeNoise-like

where ω0, ωc, ωm are the center frequency, modulator carrier frequency,

and the modulating frequency respectively

• If , then H and V transmit signals occupy same frequency band!

}]cos{[)()( 0 ttatvn

})cos{()(mod ttv mc

})cos{()()( 0 ttatv mcLSB

}]cos{[)()( 0 ttatvtH

})cos{()()( 0 ttatv mctV

02 c


Polarization diversityPolarization diversityReceive waveformsReceive waveforms

• Horizontally received signal ►

• Vertically received signal ►

• Amplitude limited horizontally received

signal ►

• Amplitude limited vertically received

signal ►

• Mixer difference output ►

► Spectrum lies between 0 and 2Spectrum lies between 0 and 2δωδω

• Mixer sum output ►

► Spectrum isSpectrum is ALWAYSALWAYS centered around centered around ωωcc !!! !!!

}]cos{[)()( 0 HHrH ttaAtv

}]cos{[)()( 0 VmcVrV ttaAtv

}]cos{[~

)(~0 HrH tAtv

}]cos{[~

)(~0 VmcVrV tAtv

}]22cos{[)( 0 HVmcdiff tBtv

}]cos{[)( HVmcsum tBtv


Noise like signal

White Gaussian noise

Frequency (Hz)Time (s)

Noise and noise-like signal Noise and noise-like signal comparisoncomparison


Amplitude and polarization angle of Amplitude and polarization angle of transmitted signaltransmitted signal


Temporal variation of electric field vector of the propagating composite wave

1

2

3

4

5

6

Instantaneous polarization vectorInstantaneous polarization vector


BER performance without codingBER performance without coding


BER performance with codingBER performance with coding


Phase Shift(delay)

Attenuation

AtmosphericAbsorption Rain Vegetation

Path Loss(distance)

Four factors that may cause distortion:

TransmittedSignal

Received Signal

Channel propagation issuesChannel propagation issues


Spectral efficiency issuesSpectral efficiency issues• Since independently generated noise waveforms

are uncorrelated, they can share same spectral space

• Non-interference feature is useful in MIMO-type polarimetric applications to avoid cross-polarization contamination

• In MIMO-type radar networking applications, many more users can be added when using noise waveforms compared to conventional waveforms


Noise waveform based Noise waveform based networking schemenetworking scheme

• Ultrawideband (UWB) noise used for attaining spread spectrum characteristics

• UWB noise radar is used for high-resolution covert target detection, tracking, and imaging

• Fragmented slices within noise band can be used for network communications (node to node and node to base station)

• Camouflaged communications appears “noise like” to adversary


Waterfilling waveform Waterfilling waveform optimizationoptimization

• Waterfilling optimization maximizes mutual information between input and output

• MIMO noise radar has many options available for optimization

• Waterfilling options in radar include polarization, operating frequency range, transmit bandwidth (resolution), spectral shaping


Waterfilling examples in radarWaterfilling examples in radar

• FOPEN applications: Higher signal losses through foliage for vertical polarization (due to vertically oriented trees) may imply the need for diverting larger fraction of transmit power to horizontal polarization

• Imaging applications: Higher bandwidth can be used to achieve better resolution from aspect locations where higher resolution is necessary to image finer identifying features of the target, while lower bandwidth (thus better spectrum usage) may be used from aspect locations where finer features may be concealed in the shadow region


Adaptive beamformingAdaptive beamforming

• Adaptive beamforming has been suggested for sensor networks

• Individual nodes respond to commands from base station and coordinate their transmissions to accomplish coherent beamforming

• MIMO radar can greatly benefit from this approach


Adaptive beamforming Adaptive beamforming examples in noise radarexamples in noise radar

• Noise radar nodes can receive “pings” from base station through the covert spectrally fragmented bands

• Standard approach would be an incoherent beamforming scheme since different noise waveforms are uncorrelated and phase synchronization is not possible

• Incoherent beamforming may only improve received power advantage by a factor of N instead of N2

• Possible to achieve coherent beamforming if pseudorandom noise waveform is used at each node


Radar tagsRadar tags

• Radar tag is a wireless device that can embed information into radar data acquisition by receiving radar pulses, modifying and coding these, and retransmitting them back to the radar

• Backscatter modulation is primarily used in sensor networks to interrogate remote devices


Applications of radar tags in Applications of radar tags in noise radarnoise radar

• Simultaneous “tagging” by each noise radar will not cross-pollinate other noise radars due to uncorrelated nature of the transmissions

• Radar tag can be designed with specific frequency dependence to be adaptive to environment conditions as viewed by each node

• Radar tags can also be used to covertly communicate information about target from one radar node to another


Noise radar networking advantagesNoise radar networking advantages• UWB Noise Radar Technology

– Noise-like transmissions for covert operations

– Large signal bandwidth, hence excellent range resolution

– LPI/LPD, anti-jam, and interference-resistant characteristics

– Efficient use of the frequency spectrum

– Low cost and compact

• Ad hoc Sensor Networks– Deployed inside or around

scene of interest– Low-cost, low-power,

untethered, multi-functional sensing devices

– Data-processing and communication

– Powerful protocol stack– Fault-tolerant and scalable– Application dependent

AmplifierNoise Signal Generator

I/Q Correlation Receiver

Antennas

VVI Q

Variable Delay Line


Proposed netted MIMO noise Proposed netted MIMO noise radar systemradar system


Possible field implementationPossible field implementation


Features of proposed systemFeatures of proposed system• It has LPI/LPD characteristics for detection,

tracking, and imaging• It can be used for covert communications and

signaling• It is based on a self-organized network-centric

architecture• The network can be used for both high and low

data rate applications• Network possesses high spectral efficiency


Combat Identification (Combat ID)Combat Identification (Combat ID)

• About 3-5% fatalities in war are due to friendly forces mistakenly targeting military targets of friendly forces (called fratricide)

• Problem is exacerbated due to adverse environmental conditions (fog/rain), harsh ground clutter, multitude of benign-looking target types (cars, etc.), crowded EM spectrum, and need to remain covert/LPD/anti-jam

• Solution requires multiple disciplines, such as sensing, communications, networking, image processing, fuzzy logic, information management, and decision sciences

Take Aways from the Combat Identification Systems Conference (CISC) held in Portsmouth, VA, May 23-26, 2005


Combat ID definedCombat ID defined

“The process of attaining an accurate characterization of detected objects in the

joint battlespace to the extent that high confidence, timely application of tactical military options and weapons resources

can occur”


Combat ID approachesCombat ID approaches

• Thermal signatures

• RF tags on vehicle

• Dynamic optical tags (DOTs) using lasers

• Millimeter wave cooperative transponder

• Microwave long range RF tags

• Digital radio frequency tags (DRAFTs)


Noise radar RF tag solution to Noise radar RF tag solution to Combat IDCombat ID

• High spectral efficiency for dense usage

• Covert operation

• LPD capability

• Anti-jam capability

• Adaptable and diverse waveform features

Questions ?Questions ?

Thank YouThank You

sensing and communications using ultrawideband random noise waveforms professor ram m. narayanan...

Documents

afosr program review

distributed frequency

uncorrelatedcenter frequency

linearstepped frequency

timefrequency equivalence

covertnoise waveforms

digital formatsnoise

communications systems