Decentralized Movement Analysis - Computational Movement Analysis - page 75

Decentralized spatial computing. Duckham's recent textbook on “Decentralized
Spatial Computing—Foundations of Geosensor networks” offers an accessible and
comprehensive introductory overview on various aspects of DeSC, including many
aspects not covered in this topic (Duckham 2012 ). Beyond movement analysis, DeSC
presents a new way of processing spatio-temporal information, contrasting to con-
ventional spatial computing in omniscient centralized spatial information systems
and databases.
Movement in decentralized spatial information systems. There are more and more
integrated spatial systems consisting of mobile networks of computing and sensing
platforms: Examples include robot swarms (McLurkin 2008 ), vehicular ad-hoc net-
works VANETs (Kosch et al. 2006 ), mobile phone networks (Ahas et al. 2010 ),
emergency dispatch (Kim et al. 2008 ), vehicle navigation, or fleet management
applications (Arampatzis et al. 2005 ). However, in most systems the conventional
centralized approach for processing and analyzing data prevails.
Decentralized movement analysis principles. There are, however, exceptions aim-
ing at explicitly decentralized ways of analyzing movement that are relevant in
the context of this chapter. Shared-ride trip planning systems, for example, are
proposed to function on a peer-to-peer basis, requiring a certain degree of in-
network data processing for allocating passengers to rides (Dillenburg et al. 2002 ;
Winter and Nittel 2006 ;Wuetal. 2007 ). Other examples can be found in the field of
robotics, where minimalist robotic swarms are tasked for inspection, maintenance,
and repair (Correll and Martinoli 2006 ; McLurkin 2008 ). Here, the system relies
on roaming mobile nodes addressing a task in a collaborative manner. The work
by Grossglauser and colleagues on mobility diffusion played a pivotal role for the
development of algorithms for decentralized flock detection (Grossglauser and Tse
2002 ; Grossglauser and Vetterli 2006 ). Aiming at improved communication in net-
works of roaming sensor nodes, these authors propose communication and routing
protocols involving (i) the maintenance of local databases that (ii) successively refine
their knowledge while moving and through using diffused information. Both con-
cepts are found in a similar way in Laube et al. ( P12 . 2011 ) adapted for the task of
decentralized movement pattern mining.
Finally, the reader is directed to further neighboring areas that may offer rele-
vant concepts and principles for decentralized movement analysis. First, there is a
large body of literature on the related but different topic of distributed data mining
in peer-to-peer networks (Datta et al. 2006 ; Kargupta and Chan 2000 ). Note that
distributed computing is less constrained than decentralized computing, such that
the cooperating systems also synchronously address a computing task but there may
be a controlling system part that has access to the entire system state. Second, the
textbook by Giannotti and Pedreschi ( 2008 ) offers an access point for privacy issues.
Apart from technical questions around privacy, the following articles also investi-
gate ethical issues and reflect on lessons learned after a decade of LBS (Dobson and
Fisher 2003 ; Uteck 2009 ; Nouwt 2008 ).