08-12-2012, 02:23 PM
Passive Sonar Fusion for Submarine C2 Systems
passive sonar.pdf (Size: 39.46 KB / Downloads: 36)
Abstract
The most important sensors for gathering target
information onboard a submarine are passive sonars. Problems
concerning fusion of these passive sonars are discussed.
Three typical passive sonars - passive noise sonar,
passive ranging sonar and acoustic pulse surveillance sonar
– are supposed to constitute a passive sonar system for data
fusion. This paper is concerned mainly with problems of
significance in system development, such as tactical application
background, special fusion techniques and own-ship
maneuver considerations.
Introduction
For tactic reasons, passive sonars are considered to be
the most important sensors onboard a modern submarine,
for which stealth is vital. Basic submarine underwater
operations, such as surveillance, search, detection
and tracking, are usually guided by passive
sonars. Almost all of modern passive sonars are capable
of processing multiple targets. They can detect,
sort, track, record and display many targets simultaneously.
When several such passive sonars are introduced
on the same platform to form a multisensor
system, fusion techniques are needed to handle this
multisensor multitarget problem. This is the task of a
unit known as fusion center, which is part of the command
and control (C2) system. Fusion center receives
and processes the multitarget information from the
sensors. The information received is usually in large
amount, of miscellaneous type, inaccurate and could
even be misleading. The output of the center is more
concise, more accurate and more meaningful tactically.
Passive Sonars
Passive noise sonar is the fundamental sensor of a
submarine. It serves both as search sensor and as attack
sensor. For positional information, noise sonar
provides the angle-of-arrival (azimuth angle, or bearing)
measurement of an acoustic source. This bearing
information is the basic information source for a submarine.
Needless to say, a comprehensive modern
passive noise sonar can provide much more information
than bearing. The accuracy of bearing measurements
is relatively good. Under some disadvantageous
conditions, however, such as in shallow water,
high water temperature, complex sea current, other
sudden changes in the underwater acoustic transmission
media, the measurement error can grow significantly.
Tactical Background
The most typical scenario of a multitarget engagement
is a submarine versus military force formation (battle
group) case. In this case, the targets are formatively
scattered. Since the movability (speed) of a marine
formation is limited and the separations between the
targets are usually large enough (compared with air
battle groups), it is quite often true that the first sensor
contact involves only one target (and most likely made
by the passive noise sonar). Gradually, as the formation
gets closer, other targets enter the sight of the
sensors, also caught by noise sonar first.
Single-Sensor Multitarget Processing
It is essential to the fusion system that each sensor
processes its multitarget positional information effectively.
The prerequisite of excellent performance of
any fusion system is that each single sensor can provide
well-sorted multitarget information within its
own domain. The most important positional information
passive sonars can get is target bearing sequences.
Therefore, the fusion problem is usually bearing-tobearing
fusion or bearing-to-track fusion. There is no
ideal tool for such fusion problems, although many
powerful techniques are available, which are, however,
more suitable for track-to-track fusion problems.
In view of this, single-sensor processing is particularly
important.
Smoothing and TMA
For a conformed sequence or track, further processing
like measurement sequence smoothing and target motion
analysis (TMA) can be done to improve the association
result. However, it is not necessarily conducted
at this stage. With more processed information
available, smoothing and TMA may be done more
effectively in the fusion center. The fact that bearingsonly
TMA is difficult and time consuming due to poor
observability [4] makes it probably better to handle it
in the fusion center. That is why the corresponding
boxes of these two parts in Fig. 2 are drawn in dashed
lines.
Gate Adjustment
With more and more information poured in, the picture
becomes clearer and clearer. It is very natural that
the association gate, usually the gate size only, should
be adjusted, although the shape also can be changed.
The size can be reduced gradually, i.e., step by step. It
can also be reduced periodically. Sometimes it needs
to be enlarged when a normal association fails. Albeit
seemingly easy, this problem can be troublesome. In
practice, however, to determine when and how to adjust
the associate gate is a problem of more engineering
than theoretical. So engineering tools, such as
simulation and trial and error, are always available and
are powerful weapons for fighting against this problem.
Multisensor Fusion
Multisensor fusion is the fusion center’s task. Because
the input data from each sensor may be bearing sequences
or tracks, three possible fusion forms exist:
bearing-to-bearing, bearing-to-track and track-to-track
fusion. Which form the fusion center should take depends
on the type of data it can get. If bearings-only
TMA is not done at the sensor level, which means
noise sonar and surveillance sonar can not provide
track data, then track-to-track fusion is not possible in
this case, because only ranging sonar can provide track
data. Even if bearings-only TMA is conducted at the
sensor level, track-to-track fusion is not the only fusion
form.
Conclusion
Passive sonar fusion is the basic and key component of
submarine sensor fusion. There are many distinctive
features in such a fusion system that need to be properly
treated. Only some major aspects have been presented.
Several problem-solving principles for system
development have also been discussed. It should be
emphasized that a modern passive sonar system could
be more complex than the model system used in this
paper [8]. The system may include more passive sonars,
and they may be more diversified. The structure
of the system itself may be quite different. In some
systems, the sonars are completely independent.
There is no information channel at the sensor level.
Some other systems, however, are highly synthesized.
All their component sonars are connected and organized
by data buses, which means the systems themselves
are distributed.