Information Technology Reference
In-Depth Information
Fig. 4.1
A geosensor
network consisting of sensor
nodes and edges of the
communication graph, given
a communication distance
c
.
The nodes measure
temperature. Nodes at the
boundary of the hot area
(
dashed line
) are shown in
black
, other nodes in
white
1
2
4
c
3
5
7
6
11
10
9
8
12
13
14
15
20
16
17
18
19
databases providing base data. However, in decentralized systems as a special case
of distributed systems, no single component knows the entire system state (Lynch
1996
). In that sense, geosensor networks are decentralized systems as nodes spread
in space, sense their local environment, but must collaborate in order to generate
the complete picture of a monitored geographic phenomenon. To this end, Duckham
(
2012
, p. 16) defines
decentralized spatial computing
(DeSC) as “the study of the
decentralized algorithms, data structures, and technologies for computing with spa-
tial information.”
The reasons explaining the need for decentralized spatial computing are manifold
(
P8
. Laube and Duckham
2009
; Duckham
2012
).
•
Information overload
. Geosensor networks may consist of thousands of sensors
sampling geospace at very fine temporal granularities. Here, decentralized spatial
computing helps managing the potentially very large and highly autocorrelated
data volumes generated in geosensor networks by filtering and processing the data
in the network.
•
Scalability
. As networks scale from hundreds to thousands and potentially millions
of nodes, centralized control of the system becomes impossible. Since decentral-
ized systems are controlled through interactions between individuals, adding more
nodes to the systems is simple, as the controlling rules can remain the same.
•
Sensor/actuator networks
. Information generated by a sensor node may often be
required by other nodes close by (e.g., controlling irrigation through a humidity
sensor network). What is sensed locally, matters locally. Removing information
from the network, processing it centrally, and then returning it into the network
would present an inefficient drain of network resources.
•
Latency
. Decentralized information processing requires communication and col-
laboration, which in turn takes time. Decentralized spatial computing can decrease
the latency of the system.
•
Privacy
. Whereas centralized databases can represent a potential security
breach, decentralized protocols can ensure that no single component can accu-
mulate knowledge about any individual, hence privacy can be protected (
P11
.
Laube et al.
2010
).
