Go to page

Bibliographic Metadata


We study the problem of gathering data and aggregation in decentralized, heterogeneous sensor networks for reliable communication. In many application scenarios, we have sensors with small energy budget and size in the network. Wildlife monitoring is one of many examples within the Internet of Things research community. To improve communication reliability and energy-efficiency of the network in such applications, macro-diversity has been employed on the data samples received by multiple sensor nodes in the network. Thus resulting in the reduction of transmission failures and further avoiding costly retransmissions. In recent times, macro-diversity techniques have been proposed which uses a distributed sensor network as an antenna array at the receiver end. These techniques primarily need the sensor nodes to forward the data samples to the sink node, to apply different diversity combining techniques. The process of forwarding the data samples from all the ground node at all times in the network incurs a huge cost. We present two algorithms, a cluster and a tree-based one, that help to reduce the data transfers in the network by pushing the aggregation process near to the point of transmission within the network. Sensor nodes within the network act as an aggregator and apply diversity combining technique on the received samples from multiple receivers rather than a centralized sink node. In an extensive set of simulations, we show that our algorithms substantially outperform naïve centralized solution and also depending upon the topology, cluster-based and tree-based algorithms outperform each other in terms of time delay and energy footprint.