Data aggregation is an important requirement in some types of wireless sensor network (WSN) applications, where generated data packets are delivered to a base station only after in-network data aggregation. Data aggregation in WSNs mitigates data traffic congestion and reduces energy expenditure. The issue of spatial data stream isolation in static routing hinders the effectiveness of data aggregation. In this paper, the authors propose attribute-aware data aggregation (ADA) based on packet attribute identifiers, a combined data aggregation and dynamic routing protocol for heterogeneous WSNs. The performance evaluation in simulation experiments validates the spatial convergence of same-attribute packets, which improves data aggregation performance in the network. The protocol is also shown to be scalable for large-scale networks.
The proposed ADA protocol consists of a packet-driven timing control algorithm and a potential-based dynamic routing (PBDR) protocol. Ant colony-based potential routing is used where, like ants, packets with the same attributes attract each other and converge. The authors propose PBDR, which combines depth potential (from sink) and attribute-based pheromone potential, to design a hybrid potential field. They also propose a packet-driven adaptive timing control algorithm, which enables the same-attribute packets to meet in time for proper data aggregation. For the same-attribute packets, the nodes maintain a timer in queue. When an attribute’s timer fires, the corresponding attribute’s data aggregation is performed. The performance of the proposed ADA protocol is evaluated in a TOSSIM sensor network simulator. ADA performance is compared with shortest-path routing (SPT), shortest-path tree (CT), dominating set (DDS), and directed tree (WCDS). In the simulation setup with 999 sensors, ADA performed much better in terms of normalized queue length, denoting better data aggregation. ADA is also shown to be scalable for large-scale networks and applicable in mobile scenarios, such as tracking mobile agents.
Overall, the idea of in-network aggregation of data packets based on packet attributes (applications) in heterogeneous WSNs is relatively novel. The paper is well written, and is recommended for readers interested in data aggregation protocols, especially in heterogeneous WSNs.