Ant-Based Delay-Sensitive Query Processing for Wireless Sensor Networks

The technologies of wireless sensor networks (WSNs) have been developed rapidly in the most recent years. As WSNs are characterized by data-centric storage and routing under most circumstances, any application involved requires data management and processing technologies, especially the query processing mechanisms. The WSNs are often deployed in diverse application specific contexts, which can be treated as distributed databases. The event-involved responses can be obtained by issuing queries to this kind of database. Certain of the applications with real-time requirement have tight constraints on query delay. However, the existing query algorithms are not satisfied with them. The problem we focus on can be simply described as: once an event has happened, how to optimize the replication energy cost due to storing the event data from the event source to multiple positions and how to optimize the searching energy cost and query delay due to finding one replica of the event and copying the event data on demand if subsequent queries for the event are performed. Ant algorithms adopt a group of intelligent agents called ants to carry out local search, using the historical information obtained to guide the ongoing search and having characteristics of collective intelligence, it can find better routing paths without knowing the global information of the network. Aiming at the real-time applications, an ant-based delay-sensitive query processing algorithm (ADSQP) taking advantage of ant colony optimization with the characteristics of self-organization and positive feedback is proposed, in which the priority-based multiple-ring storage scheme and the ant-based distributed search mechanism are adopted to improve the integrated performance of the energy-efficiency, delay, and query success rate. To reduce the replication cost of ADSQP, the cluster-based aggregation approach and the ring-based storage method according to the importance level of the events are proposed, in which the aggregated events with higher priority will be stored in the nodes on the farther ring to the aggregation nodes. To reduce the search cost and the query delay of ADSQP, the ant-based search mechanism for the queried events and the second replication approach for the found events are also proposed. The simulation results demonstrated that not only the performance of energy-efficiency and query success rate can be improved but also significant smaller query delay can be achieved by the ADSQP when comparing with other query processing algorithms. Moreover, the new algorithm is flexible and demands merely local information to find queried events efficiently and determine the number and allocation of replicas adaptively.