Decentralized Movement Analysis - Computational Movement Analysis - page 71

First, compensation leads to the build up of individual position histories, second these
histories are handed around through handshakes made possible through diffusion.
Whereas mobility diffusion has the advantage of potentially bridging longer gaps
between nodes with limited communication capabilities, this advantage comes at the
cost of latency and the lack of delivery guarantee. Both, Grossglauser and Vetterli
( 2006 ) and Laube et al. ( P6 . 2008 ) emphasize that the movement regime of the moving
nodes has a profound influence on the reliability of the mobility diffusion. The number
of encounters clearly depends on the density of the original node deployment and
their behavior. Only if nodes are expected to meet at all can this strategy offer an
advantage. Finally, Laube et al. ( P11 . 2010 ) explicitly investigated the influence of
different movement regimes in an application aiming at safeguarding privacy in an
LBS scenario. Here, the balance between level of privacy and quality of service
changed when replacing random walk with goal directed movement.
4.3.2.3 Mobility Privacy
In pre-internet and pre-database times privacy was naturally safeguarded through
the fragmented nature of personal information sources (Rule et al. 1980 ). In essence,
this early statement about privacy identifies the absence of a centralized database as
the best strategy for safeguarding privacy. Since decentralized systems per definition
do not require a centralized omniscient database, they naturally lend themselves
to offering privacy protecting services. For example, a buddy tracker application
could inform a user only through local communication when a friend enters his
neighborhood, explicitly hiding both the location of the user and the friend from any
centralized database. 5
Laube et al. ( P11 . 2010 ) explore possible benefits that decentralization offers for
safeguarding the privacy of LBS users. The paper studies this trade-off in a set of con-
secutive experiments simulating mobile and communicating agents roaming through
a real urban transportation network. The experiments vary the used communication
strategies (one-hop vs. multi-hop, push vs. pull services) and vary communication
ranges. The results clearly indicate that mobility privacy is a valid strategy for LBS.
For example, for the LBS query “Where is the nearest point of interest?”, most rele-
vant information naturally can be expected to reside in close proximity to the query
agent. Since increasing hops and communication range leads to diminishing returns
of increased quality of service, there is no need to decrease privacy through increased
hops and communication radius.
The paper furthermore discusses the notion of trajectory privacy compared to
conventional location privacy ( P11 . Laube et al. 2010 ). The idea here is, that for users
of ICT services it might be perfectly acceptable to disclose the odd static location fix,
5 Clearly, in most current ICT applications the system provider maintains a detailed log of the
whereabouts and activities of its customers, but from a conceptual point of view underlining the
argument of the mobility privacy opportunity such an omniscient system provider database is not a
necessity.