325
The Dell Precision T5500 computer has been ap-
plied as operator workstation. This machine has
64 bits processor Intel Xeon W5580, 12 GB of
RAM, two 300 GB capacity hard disks and graphical
card NVIDIA Quadro FX3800. Hard disks in the
workstation are operating under matrix system
RAID 1.
The Control Centre performs tasks as follows:
1 Generalization of information about current sea
surface situation. The process relies on verifica-
tion and data association collected from sensors,
which operate on Observation Post.
2 Monitoring, collecting and updating of sea sur-
face information within the scope of:
− Tracking of all sea surface targets in range of
Control Centre operation;
− Identification and classification of detected ob-
ject;
− Distinguishing between stationary and mova-
ble objects;
− Assertion of vessel entrance to areas temporar-
ily or permanently prohibited and other areas
defined by the operator;
− Assertion of vessel descent from in forcing
maritime routes; and
− Cooperation with others services.
3 Archiving and play back of recorded sea surface
situation.
Information received from radars and AIS are put
to the database. Next, these data are subjected to fu-
sion. Results of fusion are also written down to the
base. Communication and control of the sensors, the
database and the data fusion algorithms have been
implemented in the server.
Data from the base and the sensors i.e. radars,
AIS and even multi cameras system can be trans-
ferred automatically (in suitable range) to external
systems. Information from these external systems
can be received automatically, as well. Received da-
ta, such as targets tracks from radars and AIS are
written down to the base and are subject to fusion
with information obtained from local sensors of In-
tegrated Vessel Traffic Control System.
Information from all local sensors and external
systems are presented on two computer displays.
Actual positions and movement vectors of targets,
detected by radars and received from AIS devices
installed on vessels, are presented in the graphical
form on Electronic Chart Display and Information
System (ECDIS) (Weintrit 2009). The picture from
multi cameras system is presented in the separate
window (daylight or thermal camera). Archived
comprehensive sea surface situation can be playing
back in any moment.
Presented configuration of the model of integrat-
ed system and its functions can be changed. There-
fore, modifications and development of manufac-
tured vessel traffic control system are possible in
dependence on needs and requirements of a custom-
er.
2.2 Data fusion
Multi-sensor data fusion is one of the most effective
ways to solve problems of different groups, which
have common characteristic features. It uses data
from multiple sources to achieve a result which
would not be possible to obtain from a single sensor.
Data received from different sources can be associ-
ated by make use of specific procedures of signal
processing, recognition, artificial intelligence and in-
formation theory. Many methods of data fusion have
been developed and their common feature is an in-
clusion of multiple layers of data processing in the
integration process.
Fusion process used in presented system can be
divided into several main stages (Figure 4):
1 Unification of state vectors units of targets and
bringing them to single timeline.
2 Association.
3 Determination of an updated state vectors.
The need to harmonize the time, results from the
asynchronous operation of sensors. They operate
with different frequencies and can be turned on at
different times. For example the position of radar an-
tennas working in the system can be different at any
given time. Uniform period of time equal 1 second
has been adopted. It allows easy synchronization of
data received from sensors with different frequen-
cies (typically: 2, 3, 6, 10 seconds).
Association is performed using the modified
PDAF algorithm (Probabilistic Data Association Fil-
ter) (Bar-Shalom et al. 1995, Bar-Shalom et al.
2001, Krenc 2006). In the modification we assume
that there are two (or more) sources of varying quali-
ty. The associated measurement can come from two
sources, one or none. Additionally, there is one vali-
dation gate, inside of which are measurements from
both sources. The modification relay on adding new
innovation vectors v and association mass e of these
vectors:
j
e
i
e
j
j
i
i
v
ij
+
=
,
(1)
where: v - the innovation vector; e - association mass
of innovation vector; S - innovation matrix; and
i, j - indexes (i ≠ j).
In this way has been assumed that data are re-
ceived from both sensors and probability of such
hypothesis is calculated. This allows several percent
of improvement in the quality of estimation.