• Peter Finderup
  • Thomas Birk Abildgaard
  • Robertas Backys
2. semester, Datalogi, Kandidat (Kandidatuddannelse)
Wireless Sensor Networks (WSNs) are typically deployed in scenarios where sensor measurements from several distinct locations are needed. WSNs usually consist of many nodes scattered around an environment, often one humans tend to avoid. Example of hard to reach or hostile environments could be inside a glacier or upon a battlefield. A problem occurs once a WSN has been deployed and the running application needs a software update, either because errors has been found, parameters are tweaked or application improvements are needed. Collecting each sensor node by hand would be a tedious task, impractical if not impossible. Therefore, when such sensors need to be updated another approach is required.

The lifetime of a WSN highly depends of the utilisation of the energy resource available.
The energy supply of the individual sensor node is a very scarce resource, and the source of energy can seldom be replaced after deployment of the WSN.

In this report we focus on energy efficient code updates in a WSN through wireless communication. We systematic describe the Gossip-based Code Propagation protocol and implement it.

Through the specification of the GCP protocol we discover ambiguous protocol description, as well as limitations to the protocol.

We present an extension which incorporates a reliability mechanism which ensures updates can happen though communication is faulty. We also improve the utilisation of tokens which ensures a better overall load balancing in dense networks. As a conclusion, we compare our results from the GCP and our extension to reason about their performance while distributing code updates in a WSN.
Udgivelsesdato31 maj 2010
Antal sider106
Udgivende institutionDepartment of Computer Science
ID: 32182830