![](bg8.png)
loss rates for events and packets can be seen in Fig-
ures 7g and 7h respectively.
Tests with aggregation showed a quite low loss
rate compared to tests without. A cause may be the
nearly synchronous event production of the publish-
ers. This causes a temporal overload in the network,
which increases packet loss. The aggregation over-
comes this problem by combining multiple packets to
a single event. Therefore the short-term overload of
the network is circumvented and the additional colli-
sions are prevented.
This assumption is supported by the loss rate of
experiments L10D and L10AD. In this case the loss
rate of aggregation is higher, even though both ex-
periments showed a very low loss rate for packets as
well as events. The negative benefit might be caused
by the additionally delays of the forwarding, which
shifts forwarded events closer to the next published
event increasing collision probability.
7 CONCLUSIONS
In this paper we presented an extended pub-
lish/subscribe architecture to support application
specific aggregation. We described our architec-
ture based on the publish/subscribe middleware FA-
MOUSO and the embedded operating system Contiki
as well as the topic based aggregation.
To estimate the benefits of general aggregation
mechanisms in publish/subscribe systems we evalu-
ated different possible network topologies theoreti-
cally. Additionally we have done experiments on best
case topologies to evaluate the possible benefit. This
evaluation showed that in chain like topologies an
event count reduction of 32% for five nodes and 50%
for ten nodes is possible. This supports the theoret-
ical evaluation, which promises an increasing bene-
fit depending on the size of the chain. However the
theoretical benefits were higher because they omitted
loss rate. Additionally application-specific aggrega-
tion emerged as a solution for the temporal overload
of the network during concurrent event production of
close nodes. Finally we reasoned why aggregation
beneficial topologies will be increasingly likely de-
pending on the size of the WSN.
Following this work we want to evaluate the influ-
ence of an aggregation aware routing, which promises
additional benefits in non-optimal topologies. Fur-
thermore we want to extend the evaluation to other
metrics like energy and resource consumption.
REFERENCES
Chen, G. and Kotz, D. (2005). Policy-driven data dis-
semination for context-aware applications. In In Pro-
ceedings of the Third IEEE International Conference
on Pervasive Computing and Communications, page
283–289.
Dunkels, A. (2007). Poster abstract: Rime — a lightweight
layered communication stack for sensor networks.
Eugster, P. T., Felber, P. A., Guerraoui, R., and Kermarrec,
A. (2003). The many faces of publish/subscribe. ACM
Comput. Surv., 35(2):114–131.
Frischbier, S., Margara, A., Freudenreich, T., Eugster, P.,
Eyers, D., and Pietzuch, P. (2012). Asia: application-
specific integrated aggregation for publish/subscribe
middleware. In Proceedings of the Posters and Demo
Track, Middleware ’12, pages 6:1–6:2, New York, NY,
USA. ACM.
Harms, H. (2011). Sensornetzwerke im Bergwerk als
Untertage-Ortungssystem. Diploma thesis, Otto-von-
Guericke University Magdeburg.
IEEE (2011). IEEE standard for local and metropolitan area
networks–Part 15.4: Low-rate wireless personal area
networks (LR-WPANs). Technical Report 802.15.4-
2011, IEEE Computer Society.
Intanagonwiwat, C., Govindan, R., and Estrin, D. (2000).
Directed diffusion: A scalable and robust communi-
cation paradigm for sensor networks. In MOBICOM,
pages 56–67. ACM.
Madden, S. (2003). Tinydb: A declarative
database for sensor networks. website
http://telegraph.cs.berkeley.edu/tinydb/index.htm,
date: 2012-07-25.
Madden, S., Franklin, M. J., Hellerstein, J. M., and Hong,
W. (2002). Tag: a tiny aggregation service for
ad-hoc sensor networks. SIGOPS Oper. Syst. Rev.,
36(SI):131–146.
Madden, S. R., Franklin, M. J., Hellerstein, J. M., and
Hong, W. (2005). Tinydb: an acquisitional query
processing system for sensor networks. ACM Trans.
Database Syst., 30(1):122–173.
Pietschmann, L. (2013). dresden elektronik webseite.
http://www.dresden-elektronik.de.
Schulze, M. (2009). Famouso – family of adaptive mid-
dleware for autonomous sentient objects. website
http://famouso.sourceforge.net/, date: 17.07.2012.
the Contiki project (2012). Contiki: The open source op-
erating system for the internet of things. website
http://www.contiki-os.org/, date: 2012-07-17.
Zug, S., Schulze, M., Dietrich, A., and Kaiser, J. (2010).
Programming abstractions and middleware for build-
ing control systems as networks of smart sensors and
actuators. In Proceedings of Emerging Technologies
in Factory Automation (ETFA ’10), Bilbao, Spain.
WINSYS2013-InternationalConferenceonWirelessInformationNetworksandSystems
206