Info

Wireless Sensor Networks (WSNs) have been used by several kinds of urban and nature monitoring applications, being an important interface between physical and computational environments. Nodes clustering is a common technique to organize the network's traffic that consists in establishing cluster-heads in a hierarchical structure. Such structure reduces communication overhead and enables better network traffic management, improving issues such as scalability and energy efficiency.

Although current clustering protocols treat various kinds of dynamicity on the network, such as mobility or cluster-head rotations, few solutions consider the readings similarity. Considering the latter explicitly, though, would provide benefits in terms of better use of compression techniques and reactive detection of anomalous events. For maintaining similarity aware clusters, the synchronization of the cluster's average reading would allow a distributed and adaptive operation.

Hence, this project develops the Dynamic Data-aware Firefly-based Clustering (DDFC) protocol, to handle spatial similarity between nodes readings, and whose operation is based on the biological principles of fireflies to ensure distributed synchronization of the cluster's similar readings aggregations, differentiating thus from the classic use of fireflies. Such readings aggregation enable nodes to determine precisely how the clusters should be maintained in order to keep nodes with similar readings grouped together. Results show that DDFC is capable of keeping the cluster's readings aggregation synchronized, hence clustering nodes dynamically according to their similar readings. Furthermore, the protocol behaved with more stability, electing more appropriate cluster-heads.

The following results were achieved through this project:

  • A study about exiting clustering protocols that handle different kinds of dynamicity in WSNs. Such dynamicities were classified, including and focusing on those that regard dynamic nature of data readings;
  • The design and implementation of an adaptive clustering procol that considers the nodes data readings. Such protocol keeps the nodes with similar data clustered together, through a pre-established similarity function, without the need of centralized coordination and without the periodic complete restructuration of the network clusters;
  • A different usage of the biological principles of fireflies: through such principles, a syncrhonization method was defined for atemporal parameters;
  • A performance evaluation and thorough analysis of the proposed protocol, through the NS3 simulator. Hence, patterns between parameters and metrics were established, comparing the metrics with another protocol and a minor variation.

Last update on July, 2013.