ladar resolution improvement using receivers enhanced with squeezed-vacuum injection and...

2
LADAR resolution improvement using receivers enhanced with squeezed-vacuum injection and phase-sensitive amplification: erratum Zachary Dutton, 1 Jeffrey H. Shapiro, 2, * and Saikat Guha 1 1 Raytheon BBN Technologies, 10 Moulton St., Cambridge, Massachusetts 02138, USA 2 Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA * Corresponding author: [email protected] Received August 12, 2010 (Doc. ID 133183); posted August 12, 2010 (Doc. ID 133183); published September 15, 2010 We correct errors that were made in “LADAR resolution improvement using receivers with squeezed-vacuum injection and phase-sensitive amplification,” J. Opt. Soc. Am. B 27, A63–A72 (2010). © 2010 Optical Society of America OCIS codes: 270.0270, 280.3420, 270.6570. There were three minor typographical errors in [1]: Equation (7) is a pupil-plane representation of the phase-sensitive amplification (PSA) and hence should be written as E ˆ , t = GE ˆ R , t + G -1E ˆ R - , t . 1 Equation (13) should be T ˜ r f d ReTe -i2f· . 2 Below Eq. (16) the references should be to Eqs. (10) and (9). More importantly, there is a serious error in the deri- vation of the one-versus-two spatial resolution perfor- mance. In particular, Eq. (14) should be replaced with n r f 1 n r f 2 = S n d n d f 1 2 f 1 - f 2 + f 1 + f 2 , 3 and n i f 1 n i f 2 = S n d n d f 1 2 f 1 - f 2 - f 1 + f 2 . 4 As a result, the sufficient-statistic vector and its condi- tional covariance matrices that were developed in Section 3 are not correct. Here we will summarize a proper deri- vation of the sufficient-statistic vector and its conditional covariance matrices. In terms of these new quantities the minimum error-probability decision rule from Eq. (31) is valid. As we shall show, these corrections do not negate any of the conclusions drawn about the improvement in spatial resolution afforded by squeezed-vacuum injection (SVI) and PSA. Equations (8) and (9) of [1] provide the correct statisti- cal characterization of the quantum-enhanced LADAR. In the Section 3 treatment of resolution behavior, Eq. (15) of [1] continues to provide the target characterizations for the one-versus-two hypothesis test. Now, however, the whitening filter should be applied as follows to generate an appropriate sufficient statistic for minimum error- probability discrimination. With Y d f x defined as the 1D version of Eq. (10) from [1], we will use a whitening filter to obtain Y d f x = Y d f x S n d n d f x , 5 instead of Eq. (16) from [1]. In the space domain this gives y d x = df x Y d f x e i2f x x = G eff I T p dx ReTxmx - x + n d x , 6 where mx = df x ALf x S n d n d f x e i2f x x 7 is a real, even function, and n d x is a zero-mean, real- valued, white Gaussian noise process with correlation function n d x 1 n d x 2 = x 1 - x 2 . 8 In terms of y d x we construct a 3D sufficient-statistic vector, r T = r c rr s , as follows: r c = dx 4L 2 R 2 1/4 mx - 0 L + mx + 0 L 2 2 y d x , 9 Dutton et al. Vol. 27, No. 10/October 2010/J. Opt. Soc. Am. B 2007 0740-3224/10/102007-2/$15.00 © 2010 Optical Society of America

Upload: saikat

Post on 30-Sep-2016

215 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: LADAR resolution improvement using receivers enhanced with squeezed-vacuum injection and phase-sensitive amplification: erratum

T

pw

E

B(

vm

a

At3vcmvas(

ct[

Dutton et al. Vol. 27, No. 10 /October 2010 /J. Opt. Soc. Am. B 2007

LADAR resolution improvement using receiversenhanced with squeezed-vacuum injectionand phase-sensitive amplification: erratum

Zachary Dutton,1 Jeffrey H. Shapiro,2,* and Saikat Guha1

1Raytheon BBN Technologies, 10 Moulton St., Cambridge, Massachusetts 02138, USA2Research Laboratory of Electronics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA

*Corresponding author: [email protected]

Received August 12, 2010 (Doc. ID 133183);posted August 12, 2010 (Doc. ID 133183); published September 15, 2010

We correct errors that were made in “LADAR resolution improvement using receivers with squeezed-vacuuminjection and phase-sensitive amplification,” J. Opt. Soc. Am. B 27, A63–A72 (2010). © 2010 Optical Society ofAmerica

OCIS codes: 270.0270, 280.3420, 270.6570.

twapvt

i

w

ivf

v

here were three minor typographical errors in [1]:Equation (7) is a pupil-plane representation of the

hase-sensitive amplification (PSA) and hence should beritten as

E���,t� = �GER� ���,t� + �G − 1ER�†�− ��,t�. �1�

quation (13) should be

Tr�f� �� d� Re�T����e−i2�f·�. �2�

elow Eq. (16) the references should be to Eqs. (10) and9).

More importantly, there is a serious error in the deri-ation of the one-versus-two spatial resolution perfor-ance. In particular, Eq. (14) should be replaced with

�nr�f1�nr�f2� =Sndnd

�f1�

2���f1 − f2� + ��f1 + f2��, �3�

nd

�ni�f1�ni�f2� =Sndnd

�f1�

2���f1 − f2� − ��f1 + f2��. �4�

s a result, the sufficient-statistic vector and its condi-ional covariance matrices that were developed in Sectionare not correct. Here we will summarize a proper deri-

ation of the sufficient-statistic vector and its conditionalovariance matrices. In terms of these new quantities theinimum error-probability decision rule from Eq. (31) is

alid. As we shall show, these corrections do not negateny of the conclusions drawn about the improvement inpatial resolution afforded by squeezed-vacuum injectionSVI) and PSA.

Equations (8) and (9) of [1] provide the correct statisti-al characterization of the quantum-enhanced LADAR. Inhe Section 3 treatment of resolution behavior, Eq. (15) of1] continues to provide the target characterizations for

0740-3224/10/102007-2/$15.00 © 2

he one-versus-two hypothesis test. Now, however, thehitening filter should be applied as follows to generaten appropriate sufficient statistic for minimum error-robability discrimination. With Yd�fx� defined as the 1Dersion of Eq. (10) from [1], we will use a whitening filtero obtain

Yd��fx� =Yd�fx�

�Sndnd�fx�

, �5�

nstead of Eq. (16) from [1]. In the space domain this gives

yd��x� =� dfxYd��fx�ei2�fxx

=�Geff�IT�p

���� dx� Re�T�x���m��x − x�� + nd��x�,

�6�

here

m��x� =� dfx

A��Lfx�

�Sndnd�fx�

ei2�fxx �7�

s a real, even function, and nd��x� is a zero-mean, real-alued, white Gaussian noise process with correlationunction

�nd��x1�nd��x2� = ��x1 − x2�. �8�

In terms of yd��x� we construct a 3D sufficient-statisticector, rT= �rc r rs�, as follows:

rc =� dx4��L�2

�R2 �1/4m��x − �0L� + m��x + �0L�

2�2yd��x�,

�9�

010 Optical Society of America

Page 2: LADAR resolution improvement using receivers enhanced with squeezed-vacuum injection and phase-sensitive amplification: erratum

Gddec

,

wa

w

wplToasorcsccnttve

2008 J. Opt. Soc. Am. B/Vol. 27, No. 10 /October 2010 Dutton et al.

r =� dx4��L�2

�R2 �1/4m��x�

2yd��x�, �10�

rs =� dx4��L�2

�R2 �1/4m��x − �0L� − m��x + �0L�

2�2yd��x�.

�11�

iven the true hypothesis, r is a zero-mean Gaussian ran-om vector and thus completely characterized by its con-itional covariance matrices, �1 and �2, under hypoth-ses H1 and H2, respectively. Moreover, straightforwardalculations yield

cbsfv

DfEhd

R

1

= �4Geff�nC1

2 + C0 + C2 �2�2Geff�nC0 + 1�C1 0

�2�2Geff�nC0 + 1�C1 �2Geff�nC0 + 1�C0 0

0 0 C0 − C2

�12�

here C0, C1, C2, and n are given by Eqs. (27)–(30) of [1],nd

�2 = �� 0

0T �Geff�n�C0 − C2� + 1��C0 − C2�� �13�

ith

� = ��Geff�n�C0 + C2� + 1��C0 + C2� �2�Geff�n�C0 + C2� + 1�C1

�2�Geff�n�C0 + C2� + 1�C1 2Geff�nC12 + C0

� . �14�

Because r is zero-mean and conditionally Gaussianith covariance matrices �1 and �2, the minimum error-robability decision rule for choosing between equallyikely hypotheses H1 and H2 is given by Eq. (31) of [1].hus, to correct our resolution behavior work from [1], wenly need account for any differences between the covari-nce matrices given above and the corresponding expres-ions from [1], viz., Eqs. (24)–(26). It turns out that thenly change in going from Eqs. (24)–(26) of [1] to the cor-ect results presented above is to replace n by n /2. Physi-ally, this means that to fix the incorrect one-versus-twopatial resolution performance reported in [1] we just in-rease n by a factor of two �3 dB� in Fig. 2 and the asso-iated discussion. In particular, because this 3 dB shift inapplies to both baseline operation and all forms of quan-

um enhancement—SVI only, PSA only, SVI plus PSA—he quantitative statements made in [1] about the one-ersus-two resolution improvement afforded by quantumnhancement are unaffected. Finally, we note that be-

ause Eqs. (8) and (9) from [1] did not need correction, andecause these equations provided the basis for the imageimulations presented in [1], those simulations are unaf-ected by the corrections described herein for the one-ersus-two spatial resolution problem.

The research reported here was supported by theARPA Quantum Sensors Program. The authors grate-

ully acknowledge B. J. Yen for pointing out the error inq. (14) of [1] and its consequences. They also appreciateis confirming the correctness of the new one-versus-twoerivation presented above.

EFERENCES1. Z. Dutton, J. H. Shapiro, and S. Guha, “LADAR resolution

improvement using receivers enhanced with squeezed-vacuum injection and phase-sensitive amplification,” J.Opt. Soc. Am. B 27, A63–A72 (2010).