Professional Documents
Culture Documents
Signal Processing For Multistatic Radar Systems: Adaptive Waveform Selection, Optimal Geometries and Pseudolinear Tracking Algorithms Do Ançay
Signal Processing For Multistatic Radar Systems: Adaptive Waveform Selection, Optimal Geometries and Pseudolinear Tracking Algorithms Do Ançay
ISBN: 978-0-12-815314-7
1. Introduction 1
1.1. Historical background 1
1.2. Purpose and scope 4
1.3. Outline of book 6
v
vi Contents
Ngoc Hung Nguyen received the B.E. degree (Hons.) in electrical and
electronic engineering from the University of Adelaide, Australia, in 2012,
and the Ph.D. degree in telecommunications from the University of South
Australia, Australia, in 2016. He is currently a Researcher with the Defence
Science and Technology Group (DST Group), Australia, and an Adjunct
Research Fellow with the University of South Australia. Before joining
DST Group in 2019, he was a Research Fellow and Lecturer with the
University of South Australia. His research interests include statistical and
adaptive signal processing, compressive sensing, and estimation theory with
emphasis on target localization and tracking, sensor management, and radar
imaging. Dr. Nguyen currently serves on the Editorial Board of Digital
Signal Processing.
Kutluyıl Doğançay received the B.S. degree with honors in electrical and
electronic engineering from Boğaziçi University, Istanbul, Turkey, in 1989,
the M.Sc. degree in communications and signal processing from Imperial
College, The University of London, UK, in 1992, and the Ph.D. degree in
telecommunications engineering from The Australian National University,
Canberra, ACT, Australia, in 1996. Since November 1999, he has been
with the School of Engineering, University of South Australia, where he is
a professor and discipline leader of electrical and mechatronic engineering.
His research interests span statistical and adaptive signal processing with ap-
plications in defence and communication systems. Dr. Doğançay received
the Best Researcher Award of School of Engineering, University of South
Australia, in 2015, and Tall Poppy Science Award of the Australian Institute
of Political Science in 2005. He was the Tutorials Chair of the IEEE Sta-
tistical Signal Processing Workshop (SSP 2014), and the Signal Processing
and Communications Program Chair of the 2007 Information, Decision
and Control Conference. He serves on the Editorial Board of Signal Pro-
cessing and the EURASIP Journal on Advances in Signal Processing. From
2009–2015 he was an elected member of the Signal Processing Theory
and Methods (SPTM) Technical Committee of the IEEE Signal Process-
ing Society. He is currently an associate member of the Sensor Array and
Multichannel (SAM) Technical Committee and a member of the IEEE
ComSoc Signal Processing for Communications and Electronics Technical
Committee. Dr. Doğançay is the EURASIP liaison for Australia.
ix
Preface
Abbreviations
2D Two-dimensional
3D Three-dimensional
AOA Angle-of-arrival
BCPLE Bias-compensated PLE
CRLB Cramér-Rao lower bound
DEKF Diffusion extended Kalman filter
EKF Extended Kalman filter
FDOA Frequency-difference-of-arrival
FIM Fisher information matrix
IMM Interacting multiple model
IV Instrumental variable
IVE IV estimator
LFM Linear frequency-modulated
LOS Line of sight
MIMO Multiple-input multiple-output
MLE Maximum likelihood estimator
MSE Mean squared error
PDA Probabilistic data association
PLE Pseudolinear estimator
RMSE Root mean squared error
Rx Receiver
TB Time-bandwidth
TDOA Time-difference-of-arrival
TLS Total least-squares
TMA Target motion analysis
TOA Time-of-arrival
Tx Transmitter
UAV Unmanned aerial vehicle
WIVE Weighted instrumental variable estimator
Symbols
Rn Set of real vectors of dimension n
x A scalar
x A vector
X A matrix
I N ×N or I An identity matrix of dimension N × N or of appropriate dimension
0N ×M or 0 Zero matrix of dimension of N × M or of appropriate dimension
1N ×M or 1 A matrix of ones of dimension N × M or of appropriate dimension
x(i) The ith element of vector x
xiii
xiv List of Abbreviations and Symbols
Introduction
Contents
1.1. Historical background 1
1.2. Purpose and scope 4
1.3. Outline of book 6
Adaptive waveform
selection
Background and purpose
Waveform diversity is the concept of dynamically adapting the radar wave-
form to optimize the radar performance and confront the nonstationarity
and uncertainty of the environment [8,18]. Waveform diversity has been
exploited across different aspects of radar signal processing from detection,
tracking to classification (see e.g., [8,18] and the references therein). Various
waveform parameters can be used for waveform diversity, including pulse
repetition frequency, carrier frequency, bandwidth, pulse width, amplitude,
coding, spatial characteristics, polarization, jitter, and frequency shaping.
While waveform adaptation and agility have been only recently applied to
synthetic sensing systems like radar and sonar (which have less than 100
years of history), waveform diversity in fact has been being exploited in
nature by echolocating mammals like whales, dolphins and bats for millions
of years.
In target tracking, the tracker incorporates information about the tar-
get (e.g., range, Doppler and/or angle information) in a sequential manner,
while the radar-target geometry and sensing environment are constantly
changing due to radar and/or target motion. Therefore, dynamic adaptation
of waveforms is especially advantageous in target tracking applications as it
11
12 Signal Processing for Multistatic Radar Systems
where
1
τi,k = pk − t + pk − ri , (2.2a)
c
f◦ (pk − t)T vk (pk − ri )T vk
i ,k = + , (2.2b)
c pk − t pk − ri
py,k − ry,i
θi,k = tan−1 . (2.2c)
px,k − rx,i
Here, f◦ is the transmission frequency, c is the speed of signal propagation,
tan−1 (·) denotes the 4-quadrant arctangent, and · is the Euclidean norm.
Waveform selection for multistatic tracking of a maneuvering target 15
In (2.1), eτ,i,k , e,i,k and eθ,i,k are measurement errors modeled as zero-mean
Gaussian random variables.
Since the radar measurements can achieve their CRLBs approximately
under mild noise conditions, we follow [28–31,36] to assume that the vari-
ances of radar measurement errors eτ,i,k , e,i,k and eθ,i,k can be approximated
by their corresponding CRLBs. It is important to emphasize that the radar
measurement errors and CRLBs are known to be dependent on the radar
transmitted waveform [19]. This dependence is discussed further in the next
section, where the CRLBs for time delay and Doppler shift are explicitly
expressed as functions of transmitted waveform parameters.
Here, τA and A denote the actual time delay and Doppler shift, respectively,
while τH and H denote the hypothesized time delay and Doppler shift,
respectively. The ambiguity function (τH , H ) is maximized when τH =
τA and H = A . Here, we assume that the narrowband condition is satisfied,
16 Signal Processing for Multistatic Radar Systems
where η denotes the SNR at the receiver. Note that, consistent with the
fact that the monostatic and bistatic ambiguity functions have the same
form with respect to time delay and Doppler shift, the Fisher information
matrix {τ,} and the CRLB matrix C{τ,} for time delay and Doppler shift
given in (2.4) hold for both monostatic and bistatic cases.
Since the sharpness of the ambiguity function is determined by the
transmitted waveform as seen in (2.3), the Fisher information matrix {τ,}
and thus the CRLB matrix C{τ,} are also dependent on the transmitted
waveform. Therefore, the CRLB matrix C{τ,} can be written explicitly as
a function of the SNR at the receiver as well as the transmitted waveform
parameters ψ :
σθ2
Cθ (η) = . (2.6)
η
Although these results were originally derived for monostatic radar, they
also hold for bistatic radar because the operation of bistatic radar is basically
Waveform selection for multistatic tracking of a maneuvering target 17
the same as that of monostatic radar in measuring time delay, Doppler shift
and angle-of-arrival.
Since the CRLB for angle-of-arrival measurement is independent of the
CRLBs for time delay and Doppler shift measurements, the overall CRLB
matrix for time delay, Doppler shift and angle-of-arrival measurements ob-
tained at receiver i at time instant k is given by
and
⎡ ⎤
T5 T4 T3
⎢ 20 Qo Q
83 o
Q
62 o ⎥
⎢ 4 ⎥
⎢T T T ⎥
Q = ⎢ Qo Q Qo ⎥ . (2.12)
⎢ 8 32 o 2 ⎥
⎣T3 T ⎦
Q Q TQo
6 o 2 o
Nearly coordinated turn model
The state vector is xk = [pTk , vTk ]T . The transition matrix F is given by
⎡ sin(ωT ) 1 − cos(ωT ) ⎤
1 0 −
⎢ ω ω ⎥
⎢ 1 − cos(ωT ) sin(ωT ) ⎥
⎢ ⎥
F = ⎢0 1 ⎥ (2.13)
⎢ ω ω ⎥
⎣0 0 cos(ωT ) − sin(ωT ) ⎦
0 0 sin(ωT ) cos(ωT )
⎡ ⎤
Q11 Q12 Q13 Q14
⎢Q Q24 ⎥
⎢ 21 Q22 Q23 ⎥
Q = q⎢ ⎥ (2.14)
⎣Q31 Q32 Q33 Q34 ⎦
Q41 Q42 Q43 Q44
with
2 (ωT − sin(ωT ))
Q11 = Q22 = (2.15a)
ω3
Q12 = Q21 = Q34 = Q43 = 0 (2.15b)
1 − cos(ωT )
Q13 = Q24 = Q31 = Q42 = 2
(2.15c)
ω
ωT − sin(ωT )
Q14 = Q41 = (2.15d)
ω2
ωT − sin(ωT )
Q23 = Q32 = − (2.15e)
ω2
Q33 = Q44 = T . (2.15f)
Multiple models
In practice, a single dynamic model cannot always capture the real complex
motion of the target because the target may move straight with a constant
velocity, make turns or accelerate. In this sense, it is desirable to use the
IMM algorithm [23,24,35,43] to handle multiple models of target dynamics
simultaneously. In this chapter, we let M denote the number of models
used to characterize the target dynamics, and F m and Qm denote the state
transition matrix and process noise covariance matrix of the mth dynamic
model.
where
Stacking d̃i,k , ρ̃i,k and θ̃i,k together forms a measurement vector at receiver i
at time instant k as
ỹi,k = [d̃i,k , ρ̃i,k , θ̃i,k ]T = yi,k + ei,k = [di,k , ρi,k , θi,k ]T + [ed,i,k , eρ,i,k , eθ,i,k ]T .
(2.18)
By stacking ỹ1,k , . . . , ỹN ,k from all receivers, the overall measurement vector
at time instant k is given by
where the covariance matrix Ri,k at each receiver is computed from the
corresponding CRLB matrix Ci,k as
the EKF can produce a good tracking performance thanks to the advantage
of spatial diversity given by multistatic radar. The procedure for calculating
the EKF estimates for the mth target dynamic model is given by
where x̂m,k|j and Pm,k|j are the state estimate and error covariance,
respectively, at time instant k given measurements through time in-
stants, 0, 1, . . . , j, and H m,k+1 = H k+1 (x̂m,k+1|k ) is the Jacobian matrix H k+1
of yk+1 with respect to xk+1 evaluated at x̂m,k+1|k . The expression for the
Jacobian matrix H k+1 is provided in Section 2.8 (Appendix).
To track a maneuvering target, we employ the IMM algorithm, which
can incorporate multiple models of target dynamics. The IMM algorithm
is a technique of combining multiple state hypotheses by running multiple
filters in parallel to obtain a better state estimate for targets with changing
dynamics [23,24,35,43]. Specifically, the IMM algorithm treats the dynamic
motion of the target as multiple switching models:
where
where
m = arg max trace Pm,k+1|k+1 . (2.34)
m
24 Signal Processing for Multistatic Radar Systems
M
x̂IMM,k+1|k = μm,k+1|k x̂m,k+1|k , (2.35a)
m=1
PIMM,k+1|k
M
= μm,k+1|k Pm,k+1|k+1 +(x̂m,k+1|k −x̂IMM,k+1|k )(x̂m,k+1|k −x̂IMM,k+1|k )T .
m=1
(2.35b)
Note that the error covariance of filtered IMM estimate, PIMM,k+1|k+1 , can-
not be used because the measurement vector ỹk+1 is not available during
the waveform selection step at time instant k.
By denoting Pk+1|k+1 as the covariance matrix chosen from the three
above covariance matrices for using in adaptive waveform selection, the
waveform optimization problem is defined as
opt
ψ k+1 = arg min trace Pk+1|k+1 (ψ) , (2.36)
ψ∈
1/4
1 1
s̃(t) = exp − − j2π b t2 (2.37)
πλ2 2λ2
The track initialization is set to x̂1,0|−1 = x̂2,0|−1 = x̂0|−1 which is drawn from
a normal distribution with mean x0 and covariance P0|−1 = diag(10002 ,
10002 , 502 , 502 ).
For comparison purposes, the root mean squared error (RMSE) of the
target state estimate x̂IMM,k|k is computed from NMC = 2000 Monte Carlo
simulation runs using
NMC
1/2
1 2
n n
RMSEk = x̂IMM,k|k − xk (2.42)
NMC n=1
Figure 2.5 Patterns of selected waveform parameters for different adaptive waveform
selection schemes. (A) Minimizing combined state covariance; (B) Minimizing most-
likely state covariance; (C) Minimizing minimax state covariance; (D) Minimizing obser-
vation covariance.
Figure 2.6 Model probabilities and traces of state covariances for the uniform-motion
and maneuvering-motion models if the waveform selection is performed based on the
minimax state covariance matrix.
use of a nearly constant velocity model with q = 10 m2 /s3 . In this case, the
most-likely, minimax and combined covariance matrices become the same
as only one target dynamic model is used. Fig. 2.7 plots the RMSE per-
formance of the adaptive waveform selection method in comparison with
the minimum and maximum TB waveforms. Although the adaptive wave-
form selection outperforms the fixed waveforms in the first ten seconds,
it performs poorly after the target turn. In addition, we also observe that
the alternating sequences of λ and F no longer occur during the track
recovery period after the target turn as in the IMM-EKF case. We can
explain this by noting the inconsistency between the RMSE of the target
state estimate and the trace of the target state error covariance matrix as
shown in Fig. 2.7. We can observe that the target state error covariance
matrix Pk|k computed by the EKF does not reflect the degradation in the
real RMSE performance arising from the event of target turn. Unfortu-
nately, the waveform selection scheme decides the waveform purely based
on this covariance matrix, and thus is not able to adapt the waveform to
cope with the variation in target dynamics. This consequently results in a
poor RMSE performance of the waveform selection scheme when a varia-
tion in target dynamics occurs. Fig. 2.8 clearly demonstrates the advantage
of the use of IMM in conjunction with adaptive waveform selection.
30 Signal Processing for Multistatic Radar Systems
Figure 2.7 Simulation results with EKF. (A) RMSE performance; (B) Target state error co-
variance; (C) Patterns of selected waveform parameters.
2.7 Summary
This chapter has considered the problem of adaptive waveform selection
for multistatic tracking of a maneuvering target. Three waveform selection
schemes were presented based on minimizing the traces of the most-likely,
minimax and combined state estimation error covariance matrices of the
IMM algorithm. The performance advantages of the presented waveform
selection schemes over conventional fixed waveforms were demonstrated
by Monte Carlo simulation examples. The selection schemes based on the
most-likely and combined covariance matrices were observed to exhibit
the best RMSE performance and thus are the algorithms of choice among
all the waveform schemes considered.
2.8 Appendix
The Jacobian matrix H k of yk with respect to xk is
H k = [H T1,k , . . . , H TN ,k ]T , (2.43)
32 Signal Processing for Multistatic Radar Systems
where
⎡ ⎤
∂ d i ,k ∂ d i ,k
⎢ ∂ px,k 0 0 ⎥
⎢ ∂ py,k ⎥
⎢ ∂ρi,k ∂ρi,k ∂ρi,k ∂ρi,k ⎥
H i ,k = ⎢
⎢ ∂ px,k
⎥ (2.44)
⎢ ∂ py,k ∂ vx,k ∂ vy,k ⎥
⎥
⎣ ∂θi,k ∂θi,k ⎦
0 0
∂ px,k ∂ py,k
for the state vector xk = [pTk , vTk , aTk ]T . The derivative terms in (2.44)
and (2.45) are given by
∂ d i ,k px,k − tx px,k − rx,i
= + , (2.46a)
∂ px,k pk − t pk − ri
∂ d i ,k py,k − ty py,k − ry,i
= + , (2.46b)
∂ py,k pk − t pk − ri
∂ρi,k (py,k − ty )(vx,k (py,k − ty ) − vy,k (px,k − tx ))
= (2.46c)
∂ px,k pk − t3
(py,k − ry,i )(vx,k (py,k − ry,i ) − vy,k (px,k − rx,i ))
+ , (2.46d)
pk − ri 3
∂ρi,k (px,k − tx )(vy,k (px,k − tx ) − vx,k (py,k − ty ))
=
∂ py,k pk − t3
(px,k − rx,i )(vy,k (px,k − rx,i ) − vx,k (py,k − ry,i ))
+ , (2.46e)
pk − ri 3
∂ρi,k px,k − tx px,k − rx,i
= + , (2.46f)
∂ vx,k pk − t pk − ri
∂ρi,k py,k − ty py,k − ry,i
= + , (2.46g)
∂ vy,k pk − t pk − ri
∂θi,k py,k − ry,i
=− , (2.46h)
∂ px,k pk − ri 2
∂θi,k px,k − rx,i
= . (2.46i)
∂ py,k pk − ri 2
CHAPTER 3
Rosabelle returned to London with her uncle and aunt, very glad to
get home again. Under ordinary conditions she would have enjoyed
herself hugely at Mürren, for she was a thorough open-air girl, and
delighted in every form of sport. But the sight of other people’s gaiety
made her sad when she was so miserable herself. Mrs. Morrice, too,
seemed very unhappy and restless during what should have been
such a festive season. Rosabelle thought that Mrs. Morrice must
have been fonder of Richard than she had believed.
The first visit she paid, even before she went to see her lover, was
to the offices of Gideon Lane. This man, with his strong resolute
face, was her only hope; she had longed to be back in London so
that she might be near him; his propinquity to her gave her a sense
of comfort.
“I don’t want to make myself a nuisance, Mr. Lane, but I simply
could not keep away,” she explained by way of greeting. “You have
not been idle during our absence, I am sure. Are you any nearer to
discovering the true criminal? Have you found out anything at all?”
It was an awkward question for the detective to reply to. A very
great deal had been discovered during the time that had elapsed
between her departure for Mürren and her return to London; startling
facts at present known only to himself and Sellars.
If she had been a hard-headed practical man instead of an
emotional girl wrought up to a pitch of almost unendurable tension by
the serious plight of her lover, he might have been disposed to make
a clean breast of it. But for the moment he dared not trust her.
Guided by her feelings, she might act impulsively and spoil all his
plans.
“I will be frank with you, Miss Sheldon, as far as I can be, as I dare
be, at this juncture. Certain things have been discovered of
considerable importance. What they are, the precise nature of them,
even a hint, I dare not indulge in for the present, not until I know
much more.”
She knew the man well by now; it was useless to attempt to shake
his determination. When he had once made up his mind, it was like
beating against an open door.
“Will you at least tell me this, to ease my suspense,” she said at
length. “What you have discovered so far, does it tell against or in
favour of Mr. Croxton?”
There was a perceptible pause before he answered. Caution was
so ingrained in the man, his habit of carefully weighing every word,
his dread of expressing an opinion before he was fully justified, had
become so deeply rooted that he could hardly ever exhibit complete
frankness. Optimism was, of course, a mood unknown to him.
“If certain nebulous suspicions which are slowly forming in my
mind turn out to be correct, the result of my discoveries, so far, is
rather in favour of Mr. Croxton.”
That was all she could get out of him for the present, all she
would, in all probability, ever get out of him till he had fully and finally
solved the Morrice mystery. It was not great comfort, but it was better
than nothing.
The next day she went to Petersham, and received a warm
welcome from her lover, whose heart had been aching for her during
those weary days of separation. Small as were the crumbs of
comfort which Lane had given her, she made the most of them, and
heartened Dick considerably by her assurance that all would come
right in the end.
“There are things about the man that irritate me—just an impulsive
woman with more heart than head—his slowness, his caution, his
dislike to speak positively; but I do believe in him, in his capacity, his
ability. If the mystery is to be solved by human agency, I am
convinced he will solve it.”
The lovers had a fairly happy day, considering the depressing
circumstances. They took a long walk through busy Kingston, over
the Thames glistening in the winter sunshine, to Hampton Court,
where they had lunch at that best of old-fashioned hotels, the Mitre.
They got back to Petersham late in the afternoon, and Richard’s
kindly old nurse had a dainty tea ready for them. And too soon the
hour of parting came and her lover put her in a taxi and kissed her
fervently as they said good-bye.
“God bless you, my darling; need I tell you how I appreciate your
faith in me? But for your visits, and I count the hours till you come, I
think I should go mad. And yet I must not reckon on them. One day
your uncle will forbid them.”
The steadfast girl smiled bravely. “We will talk of that when the
time comes, my poor old boy.”
She returned his fond caress. “Remember, Dick, whether Lane
succeeds or fails, it will make no difference to me. We are
sweethearts now, and we are going to be sweethearts till the end.”
During her long absence startling events had happened in the big,
old-fashioned house in Deanery Street.
Gideon Lane had spent a busy morning away from his office. It
was three o’clock before he found time to snatch a hasty lunch.
When he got back it was close upon four. His clerk had an urgent
message for him.
“Mr. Morrice of Deanery Street has rung up three times during your
absence, sir. The last time he left word for you to go round as soon
as you came in. He said it was of the utmost importance.”
A taxi soon conveyed the detective to the financier’s house. He
found Morrice in his room in a great state of anger and excitement.
“Another robbery, Mr. Lane, this time a small one. A bundle of
Treasury notes and a quantity of Swiss bank-notes have been
abstracted, to the value of two hundred and eighty pounds. This time
I am determined to get to the bottom of it. If you are agreeable, you
shall act for me as well as for the other parties. You have no
objection to that, I suppose?”
Lane bowed. “None at all, sir. Whoever employs me does so with
the same object—to bring home the guilt to the right person.”
There were finger-marks on the safe as before. These were duly
photographed. They were identical with the previous ones, those of
the expert safe-breaker known as “Tubby” Thomas.
And “Tubby” Thomas, as they knew beyond the possibility of
doubt, was safely locked up in Dartmoor.
CHAPTER XI
A RIFT IN THE CLOUDS
C OULD there be two men the whorls of whose fingers were the
same? Scotland Yard says impossible.
“I do not question that this is a very wonderful safe, Mr. Morrice,”
the detective remarked quietly to the financier in a subsequent
conversation, “but it is evidently not as invulnerable as you and the
makers thought it.”
“Once know the secret of its mechanism, and the rest is easy,”
retorted Morrice, a little nettled at this depreciatory reference to the
wonderful invention, to the perfecting of which he had contributed not
a little himself with his own ingenious suggestions.
He explained to Lane a few of its marvels. To begin with, it was the
only safe of its kind that had ever been manufactured. The
combinations of the times when it would open would run into
millions. Supposing you worked on tens, for instance, that is ten,
twenty, thirty, the days of the month, it would only open if both keys
were applied to the same keyhole!
The detective listened politely, but he was not very interested in
learning how the thing worked; his object was to find out if there was
anybody besides young Croxton and his employer who could have
become acquainted with the secret of its working.
“It seems to me that it is really a matter of exercising ordinary
common-sense,” observed the angry banker. “Two men alone know
the secret, myself and Richard Croxton, therefore either of us could
open it after having obtained surreptitiously a duplicate of the key
held by the other. Let us assume, for a moment that you, acting on
Croxton’s behalf, say that I was the thief, that from some sinister
motive I stole my own property. Well, you are perfectly entitled to that
opinion, as an opinion.”
“I have not expressed it,” said the detective quietly.
“I know you haven’t,” snapped Morrice. He was in a very angry
mood to-day, and inclined to let his temper run away with him. “But I
also know that gentlemen of your profession cast your net wide
when you start, in the hope of catching some very unlikely fish. Of
course, I could have opened it and cast it upon Croxton, if I were so
disposed. But where is my motive for robbing myself? I can
understand certain circumstances which might induce a man to
commit such an act, and cleverly provide a scapegoat. Men set fire
to their own warehouses to get the insurance money. Why? Because
they are on the verge of bankruptcy and that money will save them.
A desperate man might steal his own money for similar reasons, to
place it beyond the reach of his creditors, so that he should not go
forth to the world a beggar. But these motives are absent in my own
case. I am more than solvent; I don’t wish to speak in any boasting
spirit, but I have more money than I know what to do with.”
Lane thought for such a practical man of the world, and
possessing, as he did, such a clear logical brain, he was indulging in
rather superfluous observations. Besides, he had referred to one
aspect of the situation as it affected himself—the absence of
financial embarrassment. If one chose to argue with him, one could
cite from the annals of crime instances of more than one other
motive that had impelled men to commit crimes which they artfully
fixed upon innocent persons, whom for some subtle reason they
wished to remove from their path, or on whom they desired revenge.
The next words, however, showed why the banker had
volunteered such an elaborate defence of himself.
“I am therefore eliminated, at any rate to my own satisfaction.” His
tone was still angry, as if he inwardly resented Lane’s rather
lukewarm attitude. “There is only left this young man whom I have
treated as a son, whom I have loaded with benefits, whom I have
preserved from the consequences of his criminal acts, his dastardly
ingratitude to me. He alone, beyond myself, knew the secret of this
safe’s mechanism, therefore he alone could open it, unless, which of
course is possible, he employed a confederate whom he took into
his confidence. Is that common-sense or not, Mr. Lane?” he
concluded in a slightly calmer tone.
“Perfectly common-sense so far as it carries us, Mr. Morrice,” was
the detective’s judicial reply.
“So far as it carries us,” cried Morrice with a slight return of his
previous explosive manner. “I do not understand you.”
Lane smiled. It was a slightly superior smile, prompted by the
thought that these clever business men, excellent and keen as they
were in their own pursuits, did not exhibit that logical mind which is
the great equipment of a trained investigator.
“By that expression I mean simply this, that when you definitely
assume Mr. Croxton’s guilt you are acting on the presumption that
nobody but you and he knows the secret of this mechanism. Can
you prove that?”
“Of course I can’t prove it in a way that might satisfy you, but I do
know that I have never told anybody else. There is of course the
maker,” he added sarcastically. “Perhaps you are including him in
your calculations.”
“I think I will consent readily to his elimination,” replied Lane in his
quiet, not unhumorous way. “Now, Mr. Morrice, we will discuss this
matter without heat. I am now employed by you as well as Mr.
Croxton, and I have only one object in view. But I must conduct my
investigation in my own way, and I want to go a little deeper than we
have yet gone.”
Morrice was impressed by the grave authoritative manner of the
man; he showed a strong touch of that quality which we notice in
eminent judges, successful barristers and all properly qualified
members of the legal profession, a patient pursuit of facts, a strongly
developed power of deduction from whatever facts are presented, a
keen faculty of analysis.
“Now, Mr. Morrice, I shall ask you a question or two for my own
enlightenment. From the little you have explained to me, the
mechanism of this safe is extremely complicated. Did you and Mr.
Croxton carry all the details of it in your heads, or had you some
written memorandum of its working to refer to in case of a temporary
lapse of memory?”
Morrice was quick enough immediately to see the drift of that
question, and his manner changed at once. “Thank you for that
suggestion, Mr. Lane; I fear I have shown a little impatience. For all
practical purposes, we did carry it in our heads, but I have in my
possession, as you surmise, a written key to which reference could
be made in the event of our requiring an elaborate combination.”
“And that key is still in your possession?”
“Yes. I keep it in the safe in my dressing-room. I looked this
morning, and it was there.”
The detective ruminated over this latest piece of information. While
he was doing so, Morrice spoke again, with just a little hesitation, as
if he knew that what he was going to say would cut the ground a little
from under his strongly expressed theory of Croxton’s guilt.
“You ought to know the whole truth, Mr. Lane, and you shall have it
for what it is worth. This present memorandum—I will speak of it by
that name—is one that I wrote out from memory. I had an original
one, perhaps just a trifle fuller, but I lost it, that is to say I could not
find it amongst my papers, some two years ago.”
Still clinging obstinately to his theory, he added a few comments
which, needless to say, did not make much impression on his
listener, who went into possibilities and probabilities perhaps just a
trifle too elaborately for the ordinary man.
“You know how easily papers get lost or mislaid. It is as likely as
not that the original memorandum will turn up in the last place I
should expect to find it. And if it got loose and was swept up by some
careless servant, it would get into the hands of the dustman. To the
ordinary person it would, of course, be quite unintelligible.”
To this Lane simply remarked that when a paper of importance
had disappeared it was quite impossible to prophesy into whose
hands it would fall. The dustman was a comforting theory, but it was
no part of his business to adopt comforting theories that did away
with the necessity to think. If “Tubby” Thomas had not been safely
locked up at Dartmoor for the last two years, he would have been
pretty certain that by some felonious means it had come into the
possession of that accomplished safe-breaker.
His position had changed since Mr. Morrice had summoned him to
Deanery Street on the occasion of the second burglary. He was now
representing the financier as well as Richard Croxton. In a way he
was glad, for Richard Croxton was poor according to Rosabelle, and
this promised to be an expensive investigation. To Morrice money
did not matter; he would not be stopped from ascertaining the truth
by lack of funds.
But in another way it was awkward. They had already found out
about Mrs. Morrice that, in conjunction with Sir George Clayton-
Brookes, the supposed wealthy baronet, she was passing off as her
nephew a young man who had no claim to the title. In course of time
Morrice would have to be acquainted with that suspicious fact, and
whatever degree of affection the banker felt for his wife, whether he
loved her very much or hardly at all, it would be a terrible blow to
him, either to his love or pride, or both.
Lane had a long talk with Sellars over the latest development of
the Morrice mystery. The young man strongly maintained that it
greatly strengthened the presumption of Croxton’s innocence, and
although the detective, with his usual habit of caution, did not take
quite such a decided view as his more impetuous lieutenant, he
readily admitted that it told in his favour, that any man possessing
the legal mind must concede as much.
“The more I can find out about him and his habits,” Sellars
remarked, “the more it seems unlikely, although not, of course,
impossible, that he should have done this thing. As far as I can learn,
he has been in love with Miss Sheldon for years, and his life has
simply been bound up with the Morrice family. They entertained very
largely, and he always showed up at their entertainments, was at
every dinner-party they gave, just like a son of the house. He seems
to have very few young men friends, but they are all of a most
respectable type. He doesn’t gamble, he doesn’t drink. Other women
don’t come into the case, for he is hardly ever a yard away from his
lady-love. Does he seem the kind of man to get himself into a hole
out of which he can only extricate himself by robbery?”
And Lane was forced to agree that if the good report of one’s
fellows could establish innocence, young Richard Croxton was
already satisfactorily whitewashed. But of course, in the opinion of
this eminent practitioner, all this was negative evidence, not positive.
Rosabelle, who was duly informed of the loss of the original
memorandum—for Lane was at bottom a very kind-hearted man and
thought he could give the harassed girl this crumb of comfort without
jeopardizing his future action—was very jubilant. She was also
pleased that her uncle had appointed the detective to prosecute his
investigations on his behalf. It would mean that Lane would not be
hampered for money.
She went over to Petersham the next day to tell her lover what had
happened, and succeeded in infecting him with her own hopeful
spirit.
“And is Mr. Morrice still as bitter against me as ever, does he still
believe as firmly in my guilt?”
Rosabelle was not very sure of the financier’s real thoughts, but
she gave the best answer she could.
“You know, my dear old Dick, how obstinately he clings to an idea
when he has once taken it into his head, but I fancy he is a bit
shaken.”
CHAPTER XII
SIR GEORGE’S VALET