ILLUMINATION FIELD OF THE INVENTION

The present invention pertains to the field of active sensing, i.e. radars, sonars, laser radars, magnetic and thermal. The concept of matched illumination can also be used in mobile communication and image processing applications.

The concept of Matched Illumination (Ml) or adaptive waveform design provides a radical shift to design of transmitters and receivers by exploiting the difference in spectral response of clutter (undesired response) and target (desired response) to adapt the transmitter modulation and receiver design to reject the clutter and enhance the return from a target. Additionally, the adaptive matched illumination waveform design can be used to facilitate better classification capability, target imagery, etc. The present invention proposes transceiver architecture for improved target detection and classification in radars.

BACKGROUND OF THE INVENTION The research activity in matched illumination radar waveform design proliferated following the classical paper by Professor Simon Haykin, "Cognitive Radar: a way of the future" and his subsequent US patent No. US 201 1 /0084871 A1 "Cognitive Tracking Radar". Following which there has been several publications and patents filed in the field of adaptive waveform design, addressing different aspects of the concept

(http://www.dtic.mil/dtic/tr/fulltext/u2/a565420.pdf). Also with the advancement of electronics, viz. re-programmable radars (http://www.militaryaerospace.com/articles/print/volume-24/i ssue-4/special- report/programmable-radar-and-adaptive-electronic-warfare-ta ke-center- s.html), arbitrary waveform generator, and high speed processors, has fuelled a surge in the environment-aware matched illumination techniques or cognitive radar. Adaptive waveform synthesis techniques are important from the perspective of next generation radar systems, as they promise 4-5dB performance improvement in terms of target detection - To achieve similar improvement in target detection, the conventional radars require to increase their transmit powers multi-folds, which calls for a proportionate rise in the cost. Additionally these adaptive design techniques also offer higher compression gains and better ambiguity function resolutions than what is available from the conventional radars. The state-of-the-art inventions in the field limit themselves to theoretical aspects and do not address how they could be tailored for practical radar systems. Also the proposed state-of-the-art inventions are adapted only to a specific radar condition. The current invention proposes architecture to handle adaptive transmission and reception under different radar conditions.

In the prior art, the US patent No. US 3,668,702 "Adaptive Matched Filter For Radar Signal Detector in the Presence of Colored Noise " is known. This invention proposes the colored noise matched filter by evaluating the statistical behaviour of the non-white noise and modifying the detection filter characteristics accordingly. This invention presents the receive-only adaptive radar system.

In the prior art, the US patent No. US 4,003,053 "Target Adaptive Radar System" is known. This invention proposes to code the transmitted radar signal so that a hard target can be considered as part of the overall system and modifies the return signal or echo so that a radar receiver can enhance the hard target echo. This invention also thereby proposes to use narrow bandwidth video amplifier.

In the prior art, the US patent No. US 5,175,552 "Optimum Matched Illumination-Reception Radar" is known. This invention presents the eigen- based optimum matched illumination waveform that maximizes the signal-to- noise ratio of the received echo by solving the Fredholm equation. The proposed solution is same as M.R.Bell 1993. In the prior art, the US patent No. US 5,121 ,125 "Optimum Matched

Illumination Waveform Design Process" is known. This invention proposes to implement the eigen-solution matched illumination solution by combining an optimal theoretical approach to maximizing the amount of energy present in the reflected radar echo with state-of-the-art computer-generated radar cross-section (RCS) codes. This invention is based on selecting a target, generating its geometric model, predicting its impulse response and its autocorrelation function and thereby selecting from a family of eigen-solution waveform via Fredholm's equation the optimal matched illumination solution that maximizes the target echo energy. Thereby, presenting one embodiment of the so called Matched Illumination-Reception Radar (OMIR).

In the prior art, the US patent No. US 5,381 ,154 "Optimum Matched Illumination-Reception Radar For Target Classification" is known. This invention proposes to utilize the optimum eigen-based matched illumination waveform for the purpose of classification. The invention aims to classify the target by attempting to associate the observed radar echo responses with a library of known target-echo signatures. This invention proposes to modify the eigen-based matched illumination waveform to facilitate target classification from a library of target signature. This US patent demonstrates for the 2-target case and mentions the m-target identification can be applied thereof. In the prior art, the PCT patent application No. WO 2006130682 A1 and the US patent No. US 7,236,124 B2 "Radar System and Method for Reducing Clutter in a High-Clutter Environment" are known. This invention deals with a system and method for generating clutter-orthogonal transmit waveform by taking SVD of the convolution matrix from the target-distorted return samples. The invention requires to apply the quantizes the clutter- orthogonal vector to the phase modulator. The radar system may perform multiple correlations on sampled radar returns from the clutter orthogonal transmit Waveform using a family of pseudo-orthogonal Waveforms to detect a slow moving target.

In the prior art, the US patent No. US 7,720,132 B2 "Energy- Bandwidth Tradeoff And Transmit Waveform Design using Interference and Noise Whitening Method" is known. This invention proposes the SINR-based Matched Illumination Waveform along with colored noise matched filter for achieving maximum SINR at the output of the receiver. This invention also highlights how the proposed SINR-based Matched Illumination can be utilized for saving either energy or bandwidth for the same output SINR performance.

In the prior art, the US patent No. US 7,538,720 B2 "Simultaneous Savings in Bandwidth and Energy using Waveform Design in Presence of Interference and Noise" is known. This invention proposes methodologies for simultaneous savings in bandwidth and energy of the transmit signal waveform and receiver output signal based on the SINR-Matched Illumination Solution. It can be viewed as an application of the SINR-Matched Illumination in saving radar resources, viz. bandwidth and energy in conventional radar. In the prior art, the US patent application No. US 201 1 /0084871 A1

"Cognitive Tracking Radar" is known. This invention provides methods and systems relating to a cognitive tracking radar (CTR) system. The invention presents a high-level implementation of the so-called Cognitive Tracking Radar (CTR), whereby the radar determines the state of a target being tracked. Based on the immediately preceding state, the system predicts parameters and waveforms, which may be used to better illuminate the target. The next reflected return forms the basis for selecting the next transmits waveform for the purpose of target tracking.

In the prior art, the following scientific publication is also known: M.R. Bell, "Information Theory and Radar Waveform Design, " Information Theory, IEEE Transactions on (Volume:39 , Issue: 5 ), Sep 1993 - The first classical scientific publication that introduces the concept of adaptive waveform design techniques in radars and also includes the use of information theory to design waveforms for the measurement of extended radar targets exhibiting resonance phenomena. The scientific publication shows that in the case of extended radar targets, the resonance phenomenon that occurs when the transmitted radar waveform is scattered by the target can be exploited to provide a larger signal-to-noise ratio at the output of the radar receiver than would result if we simply used an arbitrary waveform, viz. chirp. The scientific publication proposes optimal waveform/receiver-filter pairs design techniques for two tasks - target detection performance and target identification.

The scientific publication describes the two approaches - SNR(Eigen value) based maximization and Mutual Information based maximization. Physically the SNR-based maximization is interpreted as putting most of the transmitted energy into exciting the mode of the target with the largest eigen value. This gives us the largest possible signal-to-noise ratio and thus the best possible target detection performance under the imposed constraints. However for target identification using maximizing mutual information, the other scattering modes may be relevant and the total transmit energy is required to be distributed in a different manner such that the mutual information between target ensemble and the received waveform is maximized. In the prior art, the following scientific publication is also known: Simon Haykin, "Cognitive Radar - A way of the future", IEEE SIGNAL PROCESSING MAGAZINE, JANUARY 2006 - In this classical scientific publication, Prof. Haykin describes Cognitive Radar as one that continuously learns about the environment through experience gained from interactions with the environment and, in a corresponding way, continually updates the receiver with relevant information on the environment.

The transmitter adjusts its illumination of the environment in an intelligent manner, taking into account such practical matters as the size of the target and its range, and consequently, making adjustments to the transmitted signal in an effective and robust manner.

The whole radar system constitutes a dynamic closed feedback loop encompassing the transmitter, environment, and receiver.

The target detection is viewed from an adaptive Bayesian filtering approach in the presence of clutter and noise.

In the prior art, the following scientific publication is also known: N.A. Goodman, "Closed-Loop Radar with Adaptively Matched Waveforms", International Conference on Electromagnetics in Advanced Applications, 2007 (ICEAA 2007) - This scientific publication details the closed loop adaptive framework that is being developed at University of Arizona. It integrates a Bayesian channel representation, matched illumination techniques, and sequential hypothesis testing. The objective is to achieve target identification from a library of known target signatures. The closed loop system updates the transmit waveform based on the Bayesian channel probability update and uses sequential hypothesis testing to detect the target from the dictionary. Two matched illumination techniques are used: SNR (Eigen) based method, and Mutual Information (Ml) based methods for customizing the transmit waveform and achieving Target Identification/Classification.

The following is also prior art -

A. SINR based matched illumination (disclosed in US patents No. US 7,720,132 B2 & 7,538,720 B2)

In the presence of strong clutter, the Radar Tx-Rx baseband model and corresponding output Signal to Interference Noise Ratio (SINR) can be written as y(t) = r(t) * x(t) * h(t) + r(t) * (x(t) * c(t)

+ z( ) where y _s (t) = r(t) * x(t) * h t) represents the target- distorted signal component h(t) * x(t) at the output of the receive filter r(t), and the E[\y _n (t) \ ^A 2 ] represents the interference-noise floor at the output of the receive filter r(t) with y _n (t) = r(t) * (x(t) * c(t) + z(t)) and c(t)& z(t) are the clutter and thermal noise components respectively.

Alternately, the SINR(t = T ₀ ) can be expressed as

2) Where #( ^" ), H(/), X(f) represent the Fourier transform of the receive filter, target spatial spectra and transmitted signal spectra. <t> _c ( ^" ) and Φ _η ( ) denotes the PSD of clutter and noise.

By Cauchy-Schwarz inequality, the maximum SINR achievable can be reduced to

E

3)

And the receive filter is of the form

4)

Now the SINR(t = T ₀ can be maximized by Lagrange's multiplier technique since the objective function is concave in \X(f) \ ² with the concave constraint on transmit waveform's energy

W/2

A \J J - J

On optimization it leads to a solution of the form

\X(0\ ² = B(0(A

Thus the optimum adaptive transmit waveform-receive filter pair is given as

(6)

SNR based matched illumination (disclosed in US Patents No. US 5,175,552 & US 5,121 ,125; In the absence or weak clutter the Tx-Rx Radar baseband model and the corresponding output SNR is defined as

y(t) = * x(t) * h(t) + r(t) * z(t) SNR«

7)

The optimum transmit waveform-receive filter that maximizes SNR (in eqn 7) is obtained by using the eigen-function solution to the Fredholm equation (as in Patent US 5,175,552 &

M.R. Bell's paper) as

K( i _> ^τ 2) ⁼ j h(t— r ₁ ) i(t— T ₂ ) dt

^max ^x SNR ( ^T l) ⁼ f K( l> ^T 2) ^X SNR ( ^τ 2)^ ^τ 2 8)

Which involves taking an SVD, and is computationally expensive 0(N ³ ).

But none of the proposed inventions in fully-adaptive radar literature offers a unified solution that is capable of working under all radar conditions and assumptions. Also none of the proposed inventions are practically- compatible and do not address how hardware implementation can handle pathological signal models that can often occur in radars.

To address the above concerns, the Inventors of the present invention propose unified transceiver architecture based on adaptive waveform design that can handle any radar environmental conditions and assumptions, specifically homogenous or inhomogeneous clutter, weak or strong clutter, flat fading extended target or frequency selective extended target, and extended target or point target. The present invention/algorithm also proposes methods to take care of the pathological signal conditions that can often arise in hardware. Additionally, the present invention derives a novel matched illumination design for SNR maximizing problem, which can be implemented in O(N) time-complexity instead of the well-known eigen-based matched illumination design for the same SNR maximizing problem, which takes 0(N ³ ) time-complexity. Thus the present invention proposes a significantly efficient, practically implementable SNR matched illumination solution.

SUMMARY OF THE INVENTION The present invention aims at solving the above-mentioned drawbacks of the prior art solutions.

The present invention is defined, in its broadest sense, as a method for enhanced joint target discovery and detection in a radar through matched illumination, using an adaptive transmit and receive filter, characterized in that, in the absence or weak clutter, the modified adaptive transmit and receive filter is given as

where R (f) , H( , X(f) represent the Fourier transform of the receive filter, target spatial spectra and transmitted signal spectra and Φ _η (/) denotes the PSD of noise.

Therefore, the method according to the present invention provides the following advantages: it allows to significantly reduce the computation.

Preferably, the received clutter power is computed by evaluating

Advantageously, for frequency bins where target spectra | H( ) | ² « 0, the adaptive transmit waveform \X(f) \ ² frequency bins are blanked-out and those frequency bins are not included in the computation, i.e. According to an embodiment, the effective bandwidth of the estimated target and clutter spectra are computed as follows

max mm e^ ^ec max {Φ _ε ( )} and the adaptive transmitter is turned off if the estimated effective bandwidth of both the target and clutter is greater than a fraction of the radar bandwidth, W, i.e.

We7fec ^et > ^{aW &} ffe > ^■

According to an embodiment, the condition for operating an adaptive waveform block is

\E\v(t)\ \ ²

where y(t) is the received signal and the quantity ^_ estimates the input SINR at the receiver input.

The present invention also concerns a system for enhanced joint target discovery and detection in a radar through matched illumination, using an adaptive transmit and receive filter, characterized in that, in the absence or weak clutter, the modified ada tive transmit and receive filter is given as

R(f = k'H ^* (f)X ^* (f)e-j ²ⁿ f ^T ° where R(f), H( ), X( ) represent the Fourier transform of the receive filter, target spatial spectra and transmitted signal spectra and Φ _η ( ) denotes the PSD of noise.

T, W are radar system parameters, viz. pulse on time and radar 5 bandwidth resp. and are known to the radar apriori.

Φ _ηη ( is the Receiver Noise Power-spectral Density. The noise PSD wrt RF and temperature is calculated once during Factory Calibration for the DUT. Therefore from the Factory Calibration step, a table of Noise PSD is available at different discrete carrier frequencies (RF) and temperatures. ! O Φ _ηη ( f° ^r the given operating RF and temperature value is calculated using the Calibration table and linear interpolation methods.

The convention followed here is a hat is used to indicate estimated Noise PSD, and double nn to indicate PSD whereas single n to indicate voltage levels. However we admit we haven't maintained the convention 15 strictly throughout the manuscript.

$ _cc ( is the estimated clutter Power Spectral Density (PSD). This parameter is estimated by Environment-Aware Processing unit (in Fig. 2). Among other methods, the PSD could be estimated by Welch's method. The operating range of the radar is divided into segments, say 1 km, and the PSD 20 is being computed every pulse for each segment until a target is detected in that segment.

H(f) is the target response (both amplitude and phase) and is estimated/computed by Environment-

Aware Processing (Fig.2). Among other methods, it can be computed by

25 Ridge-Regression and/or LASSO. It needs to be noted that the adaptive architecture is proposed here is for tracking radar, i.e. a preliminary radar target is already determined through non-adaptive/fixed transmit waveform (such as LFM), therefore the position of the target is known. The target response estimation begins once a preliminary target is detected and the

30 estimation is continued for subsequent pulses until the target is visible to the radar. BRIEF DESCRIPTION OF THE DRAWINGS

The following detailed description will be better understood with the drawings, in which:

• Figure 1 represents a flow-chart of the adaptive synthesis scheduler according to the present invention ;

• Figure 2 shows a proposed architecture of the closed loop adaptive waveform/filter synthesis in tracking mode for target imaging/classification/identification;

• Figure 3 illustrates input & output SINR performances of adaptive waveform design in the presence of strong clutter environment (XCNR=10dB), i.e. TR(9) used in simulation ;

• Figure 4 represents SINR performance for adaptive transmit-receive Design (12) compared to that of adaptive transmit-receive design of prior art document US 5,175,552 and the conventional radar, i.e. TR(15) used in simulation

• Figure 5 shows NMSE performance of extended target imagery using Tikhonov regularization with LFM and Ml with 1 pulse and 20 pulse integration ; and

• Figure 6 is an alternate flow diagram.

DETAILED DESCRIPTION OF THE EMBODIMENTS OF THE

INVENTION

In the presence of strong clutter, the Inventors of the present invention propose to use the same optimum adaptive transmit waveform-receive filter pair of eqn(6) (which is already known), which is given as +

IW)I ² =

ΦΛΟ Iff 0)1

(9)

A. SNR based matched illumination

However for absence or weak clutter, the Inventors of the present invention propose to use the following optimal waveform instead of the computationally expensive eigen-solution based approach in eqn(8). The Tx- Rx Radar baseband model and the corresponding output SNR is defined as

= r(t) * x(t) * /i(t) + r(t) * z(t)

10)

The optimum transmit waveform-receive filter that maximizes SNR (in eqn 7) is obtained by using the eigen-function solution to the Fredholm equation (as in Patent US 5,175,552 & M.R. Bell's paper), whose solution requires taking SVD, which is computationally expensive. Instead, the Inventors of the present invention propose to modify the SNR cost function by taking logarithm of the function in \X(f)\ ² . Now as is known that log (l + x) « x for low x, the objective function for low SNR doesn't change, and as is noted later in the draft at higher SNR the adaptive approach doesn't play significant role. It is correct to approximate the SNR objective function with its logarithmic equivalent as suggested in the proposed invention. The modified cost function makes sense as both the objective functions are a monotonically increasing function of \X(f) \ ² .

Again, as the objective function is jointly concave in \X(f) \ ² , the optimization problem can be solved by Lagrangian method as

L{X, \X{f) \ ² )

The Karush-Kuhn-Tucker condition for the optimality of the transmit waveform is dL ro if \X(0\ ² > 0

d \X(0\ ^{2 ~} [< 0 if \X(O\ ² = 0

12)

Then the optimal transmit waveform becomes

In the presence of clutter and additive thermal noise, the optimal adaptive transmit and receive filter is

14) In the absence or weak clutter, the modified adaptive transmit and receive filter is given as ' = ( - » )*

R(f = k'H ^* (f)X ^* (f)e-j ²ⁿ f ^T °

15)

In both the adaptive transceivers architecture the parameters A & A can be computed using gradient-descent line search algorithms, such as BFGS & L-BFGS, over the transmit energy constraint, f~JX(f) \ ² df ≤ E, where E is the transmit energy specification dictated by the system.

Based of the above known framework, the Inventors of the present invention propose the following transmitter-receiver architecture:

The above framework works on different signal conditions.

That is when the received signal has strong clutter or weak clutter and orthogonality of target & clutter spatial spectra. The adaptive transmit-receive filter pair of eqn(9) or eqn(15) are proposed to be employed depending on sensing the signal. The adaptive waveform design of (15) is used when the clutter and target spectra are orthogonal in the spatial frequency. The radar signal processor and scheduler senses the received signal and determines the type of signal condition. The adaptive transmit-receive pair is decided based on the following methods -

1 . The received clutter power is computed by evaluating w

J_V| ( ) | ² $ _cc ( )d . If the parameter value is less than a threshold η, the Inventors of the present invention transmit the modified adaptive transmit and receive with modified adaptive filter of eqn(15), else the Inventors of the present invention use the transceiver pair of eqn(9). Specifically, one of the criteria to choose between transceiver between (9) and (15) can be expressed as

w

2 _{| 1 m i 7 ¾ m j i} (TRUE use 77?(15)

w_ ^{y y i y} ' ¹ (FALSE use 7 ? (9)

(16)

TR is abbreviated for Transceiver.

When the target and clutter are orthogonal in spatial frequency, i.e. say the clutter occupies left-side of radar baseband spectra whereas target is right-shifted of the baseband spectrum with very little spectral overlap, it makes sense to ignore the clutter spectra completely while designing the adaptive waveform, as selective transmission within clutter spectra will not improve the target detection. Hence under such conditions the adaptive transmission and reception can entirely be influenced by the target and noise spectra. The Inventors of the present invention propose to categorize such scenarios by the following criteria

(TRUE use TR(lS)

[FALSE use TR(9)

The formulation of above Transceiver doesn't render itself to practically implementable sub-systems. One of the key bottlenecks is the divide by 0 issue, which can happen in computation of X( ) in both TR(6) & TR(12), precisely they can arise for frequency bins where the clutter PSD <t> _c (/) and/or target spectra \H(f) \ ² is negligible or close to

0. To overcome such pathological conditions, which can arise often in the adaptive waveform design, the Inventors of the present invention propose the following methods -

1 . For frequency bins where target spectra |H( ) | ² « 0, the adaptive transmit waveform \X(f)\ ² frequency bins are blanked-out and those frequency bins are not included in the computation, i.e.

(TRUE fe I

i ^S WW ~ ° [ FALSE fe U

(1 8)

Precisely only the frequency set n is rendered for the computation of adaptive waveform design.

2. The frequencies where the clutter spectra <t> _c ( ^" ) « 0, is taken care by adding a constant offset to the clutter PSD <t> _c ( ^" ). This constant offset is determined as some factor of the l/( _{t< rflet} ) , where E is

^W eff

the total radar energy, and W^ ^get is the effective target bandwidth. The rationale behind the choice of the constant factor is owing to the fact that the maximum radar energy would be split within target effective bandwidth, W^ _f ^raet .

From practical embedded systems stand point, it is important to turn off the adaptive waveform synthesis module when no advantages are guaranteed from adaptive transceiver to preserve power and overall radar performance. The Inventors of the present invention design methods to turn off the adaptive waveform synthesis module as follows -

1 . When there is no diversity in the target and clutter spectra, i.e. the target is point or near-point and the spatial clutter is independent or near-independent, the adaptive waveform synthesis techniques can exploit no or very little advantage than conventional radar. The Inventors of the present invention compute the effective bandwidth of the estimated target and clutter spectra as follows yclutter _ ^J -W/2

max {4> _c ( ^" )}

(19)

The Inventors of the present invention propose to turn off the adaptive transmitter if the estimated effective bandwidth of both the target and clutter is greater than a fraction of the radar bandwidth, W, i.e.

We7fe ^a c ^et > ^{aW &} > (20)

The typical values of a and β are between 0.5

to 0.75, and W is the operating bandwidth of the

radar. Both a and β values lie between (0,1 )

and are scalar. The idea here is to utilize

selective transmission even if either target

and/or clutter PSD has an effective bandwidth

less than 0.5 to 0.75 times the operating

bandwidth. If the effective bandwidth is more

than 0.5 to 0.75 times the operating bandwidth

for both target and clutter, then there is very

little likelihood that there would be any

performance incentive in using

adaptive/selective transmission over standard non-adaptive radar pulse (LFM/NLFM).

When the input SINR is less than a pre-defined SINR (computed offline based on the target estimator) and greater than a pre-defined SINR, turn off the adaptive transmitter. The lower SINR cutoff is required based on the accurateness /performance of the target estimator used, as below SINR _i0W the performance of target estimate would be too poor for the adaptive waveform to leverage any performance advantage. This can be decided offline based on the performance of the target estimator used. The SINR _h ig _h value is computed as SINR _high = J_^ _{/2 $} ' ^ df, S I N R _!ow is predefined and rather computed empirically for each system, and is dependent on the lowest SINR where a target can be detected. Typically it would lie somewhere between -10dB to -5dB.

Also above a pre-defined SINR, the optimum transmit signal almost resembles the waveform of that of LFM/conventional waveform as illustrated in the figures below. In such a case too, the adaptive waveform synthesis unit can be turned off. Thus the condition for operating the adaptive waveform block is

≤SINR high

21 ) where y(t) is the received signal and the quantity

|£[y(t)]| ²

estimates the input SINR at the receiver input.

Figure 1 represents a flow-chart of the adaptive synthesis scheduler according to the present invention. Figure 2 shows a proposed architecture of the closed loop adaptive waveform/filter synthesis in tracking mode for target imaging/classification/identification.

Figure 2 presents the overall architecture of the adaptive radar design. The additional components attributed to the idea are Adaptive Phase Modulator, Environment-aware Processing, Adaptive Synthesis Scheduler and Adaptive Correlator/Filter. The Adaptive Synthesis Scheduler (ASS) is the central piece of the architecture and implements the flowchart below, deciding the waveform-filter pair to be transmitted when a preliminary target is detected, i.e. in tracking mode. The ASS also controls the Environment- Aware Processing (EAP) unit wherein the regular clutter PSD estimation is carried out during radar non-tracking mode across all segments, and instructs EAP to switch to no estimation on the segment where target is first detected and subsequent segments where the target moves. The ASS module takes care of the entire adaptive processing chain and also takes care of the power management through turning adaptive transmissions when above criteria are not met. In the radar tracking mode, the output of the ASS is the desired transmit spectra \X(f) \ ² . The adaptive phase modulator takes in the desired transmit spectra as the input and reconstructs the time phase of the transmit signal (using phase retrieval algorithm based on the earlier

PCT patent application No. PCT/FR201 3/053034), and the data is sent to the constant envelope module for on-air transmission and the phase data is communicated back to ASS, where the receive filter is also designed and then communicated to the adaptive correlator/filter module.

Figure 3 illustrates input & output SINR performances of adaptive waveform design in the presence of strong clutter environment (XCNR=1 0dB), i.e. TR(9) used in simulation. Figure 4 represents SINR performance for adaptive transmit-receive

Design (1 2) compared to that of adaptive transmit-receive design of prior art document US 5,175,552 and the conventional radar, i.e. TR(15) used in simulation.

Figure 4 represents the output SINR vs transmit energy of the SINR- matched illumination (target, clutter and receiver noise adaption), SNR- matched illumination (target and receiver noise adaption) and LFM transmit signal. The figure depicts that both SINR and SNR based adaptive transmission presents substantial performance improvements over conventional LFM waveform.

The present invention aims to improve the detection and target imaging (Non-Cooperative Target Identification/Classification) performance in radar tracking mode. Fig. 3 & Fig. 4 represent the performance improvement for target detection. Fig. 5 here represents the NMSE performance of the target imaging process, defined as ^( - h) ^H (h - h) where h \s the estimated target response and h is the actual target impulse response. The target impulse response is estimated using Ridge Regression and the comparison is on using standard non-adaptive LFM waveform and on using SINR-MI waveform with single-pulse and 20 pulse-averaging.

Figure 5 shows NMSE performance of extended target imagery using

Tikhonov regularization with LFM and Ml with 1 pulse and 20 pulse integration

Figure 6 is an alternate flow diagram. This alternate flow-diagram could be viewed as a superset of the flow-diagram of Figure 1 by taking SINR_mid = SINRJow.

The adaptive waveform synthesis transceiver could be used either for target detection, SAR imagery and target classification applications. The adaptive waveform synthesis as highlighted above can be used to achieve pre-defined SINR/SNR or NMSE performance using much lesser transmit power and minimal additional computational load. The method for selecting non-adaptive transmission, i.e. conditions over which to schedule are estimated SINR (or radar transmission power) and target/clutter effective bandwidth. Note that the G1 and G2 equations below are used for arbitrating which adaptive waveform-receive filter pair to use for adaptive transmission. The expression for matched filtering for corresponding LFM/NLFM is as follows:

R(0 = kX ^* (f)

Where k is a constant and X( ) is the radar transmitted signal spectrum.

Expression G1 computes the received clutter power, i.e.

| ² _cc ( )d , and compares it with η (corrected in flowchart now).

The parameter η is determined from the receiver noise floor, kTWF _n , and is chosen (1 /10) to (1 /50) times kTWF _n . The rationale here is that if the received clutter power is 10-15dB lesser than the receiver noise, then the clutter power is too much subdued by the receiver thermal noise, and thus adaption based on clutter spectra is not expected to offer any performance benefits. So a waveform design which adapts based only on target and receiver noise is used, such a waveform is termed SNR matched illumination. However if the received clutter power is more than η, a waveform design which adapts based on target, receiver noise and also clutter is used. Such a waveform is termed SINR matched illumination.

Similarly G2 expression, G2 = computes the s \ \ ² d _f s $cc(j)cif

orthogonality between the target response and clutter PSD response. ?7'here is chosen to be between (1/50) to (1 /100) times /c7 F _n (receiver noise floor). Hence if the orthogonality between target spectra and clutter PSD is high, i.e. low G2 value < η' , the waveform design is based only on target and receiver noise is chosen. The idea is to maximize target detection, and in cases where the target and clutter occupy non-overlapping regions of the spectra, for example say target spectra is towards left of the baseband and the clutter is towards right, then it makes sense only to adapt based on the target spectra and the receiver noise, and ignore the clutter response in designing the adaptive waveform. Similarly on the other hand, if the clutter PSD and the target response occupy substantial overlapping regions of the spectra, a selective transmission that takes into consideration - the target, the clutter and receiver noise should be used.

The above specification, examples and drawings provide a complete description of the method according to the present invention. Since many embodiments of the invention can be made without departing from the spirit and scope of the invention, the invention resides in the claims herein after appended.