Radar Signal Processing

Ambiguity Function and Waveform Design Golay Complementary Sequences (Golay Pairs) Golay Pairs for Radar: Zero Doppler

Radar Signal Processing

Radar Problem

Transmit a waveform s(t) and analyze the radar return r(t): r(t) = hs(to )ej(to ) +n(t) h: target scattering coecient; o = 2do /c: round-trip time; o = 2fo 2v c : Doppler frequency; n(t): noise Target detection: decide between target present (h = 0) and target absent (h = 0) from the radar measurement r(t). Estimate target range d0 . Estimate target range rate (velocity) v0 .

Radar Signal Processing

Ambiguity Function
Correlate the radar return r(t) with the transmit waveform s(t). The correlator output is given by

m( o , ) =

hs(t o )s(t )ej(to ) dt + noise term

Without loss of generality, assume o = 0. Then, the receiver output is m(, ) = hA(, ) + noise term where A(, ) =

s(t)s(t )ejt dt

is called the ambiguity function of the waveform s(t).

Radar Signal Processing

Ambiguity Function
Ambiguity function A(, ) is a two-dimensional function of delay and Doppler frequency that measures the correlation between a waveform and its Doppler distorted version:

A(, ) =

s(t)s(t )ejt dt

The ambiguity function along the zero-Doppler axis ( = 0) is the autocorrelation function of the waveform:

A(, 0) =

s(t)s(t )dt = Rs ( )

Radar Signal Processing

Ambiguity Function
Example: Ambiguity function of a square pulse

Picture: Skolnik, ch. 11

Constant velocity (left) and constant range contours (right):

Pictures: Skolnik, ch. 11 Radar Signal Processing

Ambiguity Function: Properties

Symmetry: A(, ) = A(, ) Maximum value:

|A(, )| |A(0, 0)| =

|s(t)|2 dt

Volume property (Moyals Identity):

|A(, )|2 d d = |A(0, 0)|2

Pushing |A(, )|2 down in one place makes it pop out somewhere else.
Radar Signal Processing

Waveform Design
Waveform Design Problem: Design a waveform with a good ambiguity function. A point target with delay o and Doppler shift o manifests as the ambiguity function A(, o ) centered at o . For multiple point targets we have a superposition of ambiguity functions. A weak target located near a strong target can be masked by the sidelobes of the ambiguity function centered around the strong target.
Picture: Skolnik, ch. 11

Radar Signal Processing

Waveform Design
Phase coded waveform:
L1

s(t) =
=0

x( )u(t T )

The pulse shape u(t) and the chip rate T are dictated by the radar hardware. x( ) is a length-L discrete sequence (or code) that we design. Control the waveform ambiguity function by controlling the autocorrelation function of x( ). Waveform design: Design of discrete sequences with good autocorrelation properties.

Radar Signal Processing

Phase Codes with Good Autocorrelations

Frank Code

Barker Code

Golay Complementary Codes

Radar Signal Processing

Waveform Design: Zero Doppler

Suppose we wish to detect stationary targets in range. The ambiguity function along the zero-Doppler axis is the waveform autocorrelation function:

Rs ( ) =

s(t)s(t )dt
L1 L1

=
=0 m=0 L1 L1

x( )x(m)

u(t T )u(t mT )dt

=
=0 m=0 2(L1)

x( )x(m)Ru ( + (m )T )
L1

=
k=2(L1) 2(L1) =0

x( )x( k)Ru ( kT )

=
k=2(L1)

Cx (k)Ru ( kT )

Radar Signal Processing

Impulse-like Autocorrelation
Ideal waveform for resolving targets in range (no range sidelobes):
2(L1)

Rs ( ) =
k=2(L1)

Cx (k )Ru ( k T ) ( )

We do not have control over Ru ( ). Question: Can we nd the discrete sequence x( ) so that Cx (k ) is a delta function? Answer: This is not possible with a single sequence, but we can nd a pair of sequences x( ) and y ( ) so that Cx (k ) + Cy (k ) = 2Lk,0 .

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Complementary Sequences (Golay Pairs)

Denition: Two length L unimodular sequences x( ) and y ( ) are Golay complementary if the sum of their autocorrelation functions satises Cx (k ) + Cy (k ) = 2Lk,0 for all (L 1) k L 1.

Radar Signal Processing

Golay Pairs: Example

x y

Time reversal: x : 1 1 x: 1 1 1 1 1 1

1 1 1 1 1

1 1 1

If (x, y ) is a Golay pair then (x, y ), (x, y ), and (x, y ) are also Golay pairs.
Radar Signal Processing

Golay Pairs: Construction

Standard construction: Start with construction
A B A A B B 1 1 1 1

and apply the

B B A A

Example:
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

1 1

1 1

Other constructions:
Weyl-Heisenberg Construction: Howard, Calderbank, and Moran, EURASIP J. ASP 2006 Davis and Jedwab: IEEE Trans. IT 1999
Radar Signal Processing

Golay Pairs for Radar: Zero Doppler

The waveforms coded by Golay pairs x and y are transmitted over two Pulse Repetition Intervals (PRIs) T . Each return is correlated with its corresponding sequence: Cx (k ) + Cy (k ) = 2Lk,0
x y

Discrete Sequence

Coded Waveform
Radar Signal Processing

Golay Pairs for Radar: Advantage

Frank coded waveforms

Golay complementary waveforms

Weaker target is masked

Weaker target is resolved

Radar Signal Processing

Sensitivity to Doppler

Although the autocorrelation sidelobe level is zero, the ambiguity function exhibits relatively high sidelobes for nonzero Doppler. [Levanon, Radar Signals, 2004, p. 264] Asx (, ) + ej 2T Asy (, ) Why? Roughly speaking Cx (k ) + Cy (k )ej = ()k,0

Radar Signal Processing

Sensitivity to Doppler
Range Sidelobes Problem: A weak target located near a strong target can be masked by the range sidelobes of the ambiguity function centered around the strong target.

Range-Doppler image obtained with conventional pulse train

x y x y

Radar Signal Processing

References

1 2 3 4 5

M. I. Skolnik, An introduction and overview of radar, in Radar Handbook, M. I. Skolnik, Ed. New York: McGraw-Hill, 2008. M. R. Duco and B.W. Tietjen, Pulse compression radar, in Radar Handbook, M. I. Skolnik, Ed. New York: McGraw-Hill, 2008. S. D. Howard, A. R. Calderbank, and W. Moran, The nite Heisenberg-Weyl groups in radar and communications, EURASIP Journal on Applied Signal Processing, Article ID 85685, 2006. N. Levanon and E. Mozeson, Radar Signals, New York: Wiley, 2004. M. Golay, Complementary series, IRE Trans. Inform. Theory, vol. 7, no. 2, pp. 82-87, April 1961.

Radar Signal Processing