QoS supported efficient clustered query processing in large collaboration of heterogeneous sensor networks

Significant worldwide growth is witnessed in development and deployment of huge numbers of heterogeneous sensor networks. These all brings the issue of state-of-the-art federations or collaborations among such networks. Query Processing operation in such collaborative systems has major challenges: fast and scalable query processing, QoS support for query, flexible and robust collaborative system design etc. To our knowledge there has not been much work done on designing scalable and efficient query processing among the huge collaboration of sensor networks. The work EE-QPS designed a pipelined query optimization problem based on energy efficiency. But with varying demands (energy, delay, reliability etc.) of different queries, the Quality of Service (QoS) support becomes very important. Also, entirely sequential or entirely parallel query processing have problems with latency and scalability. Then a hybrid query processing scheme can have flexibility to deliver better performance for all kinds of collaborative systems. Considering all these aspects, we have proposed QoS-QPS, a QoS supported clustered Query Processing System. We have designed a flexible model for querying cost of sensor networks. The cost model is inexpensive to compute and general enough to apply. Then we propose clustered query processing technique, that utilizes a constrained graph partitioning algorithm. This whole QoS aware query processing technique delivers balanced and efficient clustering of sensor networks based on implication relationship. Comprehensive simulations study shows that our proposed scheme is better than existing techniques in compromising among different system requirements. The results also validate the efficiency, scalability and applicability of QoS-QPS. Further we have analyzed potential architectural issues and possible solutions.