A common wireless sensor network architecture?
Vlado Handziski, Andreas Köpke, Holger Karl, Adam Wolisz
Telecommunication Networks Group
Technische Universität Berlin
1 Fragmentation of research
Wireless sensor network (WSN) research is currently heavily fragmented: Independent developments
take place on both sides of the Atlantic, in many different research groups. There is not yet a common
“lingua franca” as in some other fields of networking research, where e.g. Linux on Intel processors
or the Network Simulator  have achieved such a status—with all the advantages and disadvantages of
such a predominance. In WSNs, the need for a harmonization of research efforts is particularly felt in the
Simulation environments Many different research groups use different simulation environments
for performance evaluation. Currently popular tools include NS/2, Opnet, GlomoSim, Qualnet, or Om-
NET++. None of these simulation tools perfectly meet the demands of WSNs. For example, the wireless
channel model is too simplistic and not easily changed (NS/2), not all relevant protocols are easily avail-
able (Omnet), some are commercial (Opnet, Qualnet), some are too infrequently used to easily allow
comparison of results with work of other researchers (all tools with the possible exception of NS/2).
More importantly, there is no simulation tool for which convincing models of applications or sensor
excitation are available.
Hardware platforms The overall setup of sensor node prototypes that have been developed and are
currently used by different groups for experimental research is in principle very similar. Examples for
such nodes include the Mica Motes, the nodes used by the EYES project , and nodes developed locally
at Technische Universitt Karlsruhe, ETH Zrich, Freie Universitt Berlin, or Technische Universitt Dresden.
Typical microcontrollers are the Atmel, Motorola microcontrollers or the Texas Instruments MSP 430;
usable radio modems include those by RFM,