Power Modeling and Simulation for Wireless Sensor Networks

Staff - Faculty of Informatics

Start date: 31 January 2011

End date: 1 February 2011

You are cordially invited to attend the PhD Dissertation Defense of Marcello MURA on Monday, January 31st 2011 at 11h00 in room SI-008 (Informatics building)

Abstract:

WSNs are large - possibly complex - structures made of simple nodes, that are in general required to operate without human supervision for long time intervals. The main breakthrough is due to the fact that physical data (gathered from sensors), and not digital data, are the input of the computation, therefore effectively moving the interface of the system towards the physical world. Networking plays a central role as WSNs are inherently distributed; being nodes spread in space and without any physical connection, their coordination requires particular attention. Many proposals have been presented in the literature concerning algorithms, models or heuristics for WSNs: the rationale of current research in the area is that of developing general solutions that may be used (after a proper tuning) for suitable sets of applications.

In Pervasive Computing and in WSN in particular the most challenging problem when developing applications is given by energy shortage. Nodes are normally supplied through batteries, hence a limited amount of energy is available and no breakthroughs are foreseen in a near future; when it is possible to harvest energy from the environment the amount of energy harvested impacts on the dimension and the cost of the node. For this reason power analysis and power optimization, both at node and at network level, have received great attention. In particular, definition of policies to save energy have been targeted. In this work, first of all the problem of modelling power consumption for WSNs is studied, the aim being that of defining models as far as possible related to protocols rather than to specific node technologies - therefore as far as possible general - and then to characterize such models for specific implementations. These models constitute the basis for subsequent simulation tools and for targeting node-and networklevel optimizations and enhanced energy saving policies are devised.

In particular we have worked on applying protocol level hierarchical modelling to the WSN field, models are platform independent and are later tied to one or more particular implementations through characterization. This way the same model can be used for different implementations of the same algorithm and performance comparison is natural. Our modelling framework is integrated with the capability of automatically generating executable simulator starting from models.

The proposed techniques are applied to various WSN protocols. In particular we built a complete model of the MAC layer of IEEE 802.15.4 standard and modelled the networking layer of ZigBee standard. Such models are used to analyse the power performances of such protocols and design suitable optimizations. Finally we used our technique for comparing different WSN solutions (i.e. ARBUTUS protocol and ISA 100.11a standard) for an environmental monitoring network to be deployed in a solar system planet.

Dissertation Committee:

  • Prof. Mariagiovanna Sami, Università della Svizzera italiana, Switzerland (Research Advisor)
  • Prof. Matthias Hauswirth, Università della Svizzera italiana, Switzerland (Internal Member)
  • Prof. Laura Pozzi, Università della Svizzera italiana, Switzerland (Internal Member)
  • Prof. Giuseppe Anastasi, Università di Pisa, Italy (External Member)
  • Prof. William Fornaciari, Politecnico di Milano, Italy (External Member)