Information Technology Reference
In-Depth Information
each other's local neighbor memory required for the actual deferred data mining
task. In both cases, mobility is essential as it leads to new node-node combinations
and hence a recombination of opportunities for exchanging information. Finally, the
“fish” in Both et al. (
P19
.
2013
) collaborate since nurturing the memories of the
cordons requires fish acting as data mules delivering information packages from
other near-by cordons. All of the above examples also illustrate the collaborative
nature of
mobility diffusion
and
mobility resilience
.
4.3.3.2 Local Relevance of Locally Sensed Information
In many geosensor systems information that is sensed by spatially distributed nodes
will be most relevant to other nodes in their immediate spatial vicinity. This is espe-
cially the case in
sensor/actuator networks
where the locally sensed and processed
information is also required locally by the network, for instance for a local activation
of irrigation nozzles based on local sensor readings (
P8
. Laube and Duckham
2009
;
Duckham
2012
). This principle—what is sensed locally is relevant locally—shows
up repeatedly in the research summarized in this chapter. In decentralized flock detec-
tion algorithms nodes limit communication to other nodes in their vicinity and but
not the other nodes in the network (e.g.,
P12
. Laube et al.
2011
).
In the LBS application in Laube et al. (
P11
.
2010
) this local relevance is the
basis for trading the level of privacy and the quality of service. Since local POIs are
more relevant, communication can be limited and hence (trajectory) privacy remains
protected (see also Sect.
4.3.2.3
).
4.3.3.3 Latency
Latency refers to the length of the delay between when an event occurs and when
that event is correctly detected by an algorithm (
P19
.Bothetal.
2013
; Duckham
2012
). As the title suggests, in Laube et al. (
P12
.
2011
) on “Deferred decentral-
ized movement mining” roaming nodes require a latency phase for building up their
neighborhood memory. The mobile nodes compensate for their limited spatial per-
ception range through extending their temporal perception range, that is, allow for
a latency period before the actual data mining step kicks in. Here, computation is
deferred, but it is also possible to defer communication. In Laube et al. (
P6
.
2008
;
P12
.
2011
), and in Both et al. (
P19
.
2013
) roaming nodes are given time to rearrange
and recombine communication opportunities. Latency is an implication of
mobility
diffusion
since time is required for the information to move through the network.
4.3.3.4 Separation
Another feature of decentralized systems is the separation of tasks which in conven-
tional spatial computing are typically hosted in a central and omniscient computing
platform. Apart from the obvious spatial separation of partial spatial computing tasks
that are, for example, required for computing a boundary, separation can also be of
