Information Technology Reference
In-Depth Information
The second distinction addresses the question whether the system only generates
information or whether the system allows also for querying this information. This
distinction also addresses the question where in the system the information is stored.
Often, decentralized systems will have separate algorithms for first generating infor-
mation or collating information in specific nodes in the network. The querying of this
information will be delegated to separate algorithms (Duckham
2012
). For example,
in Both et al. (
P19
.
2013
) most algorithms are concerned with the collation of infor-
mation about the flow in the network in the cordons. Specific algorithms presented in
the final section of the paper then investigate how this system would handle queries
put to the monitoring system.
In the following two decentralized movement analysis tasks illustrate two opposite
system architectures and related problems in the light of the general issues raised in
this section. The examples include the decentralized monitoring of network flows in
a cordon-structured network (mode III, Sect.
4.2.1
) and the decentralized detection
of flock movement patterns (mode IV, Sect.
4.2.2
).
4.2.1 Static Nodes Monitor Mobile Objects
Unconstrained movement is rare. People move, for example, mostly in constrained
transportation networks. Hence, infrastructure enabling, managing and monitoring
network-bound movement becomes an important source for large movement data
volumes. Examples range from cellular networks for mobile phones to electronic
ticketing for public transport (e.g. London's Oyster card or Melbourne's myki system)
and road tolling systems. In the context of this chapter such systems are interesting
since some capture information in a decentralized way, where cars are observed
when passing GSM towers or traffic cordons or commuters swiping card readers
when hopping on and off trains or buses. Collating all that information capturing
the whereabouts of agents in such systems in centralized databases may at best be
impractical, in some cases simply impossible. Hence such systems lend themselves
to decentralized spatial computing and decentralized movement analysis.
Both et al. (
P19
.
2013
) present a family of algorithms for the decentralized
monitoring of moving objects in such a transportation network augmented with
checkpoints.
3
The approach considers mobile objects that move and are tracked
on a
transportation network
, modeled as a graph where transportation edges con-
nect intersections. The moving objects are tracked whenever they pass a cordon
(or checkpoint) that are typically but not necessarily positioned at intersections.
Figure
4.2
depicts a generalized ring-shaped network with four cordons indeed posi-
tioned at the intersections. The movement in this architecture is
constrained
to a
3
In Both et al. (
P19
.
2013
) “fish” is used as a shorthand for moving objects because the work was
initiated in response to a set of problems coming out of a river health monitoring system deployed
in the Murray River, Australia, tracking real fish with RF transmitters and riverside cordons (Koehn
et al.
2008
).
