Abstract: On data-centric distributed sensor network, users often submit queries about the interested sensing region such as gathering the data in region F during a specified period of time. In order to enhance monitoring accuracy, sensor networks rely on high density deploying node and sampling frequency, so there are lots of redundant data. Therefore, guaranteeing querying data accuracy and reducing energy consumption of transmitting redundant data are an urgent problem to be resolved at present. This paper proposes an single node sensor data time slice fitting algorithm, the proposed algorithm can balance the energy consumption of node in query region; secondly, an slice model fitting algorithm, which can find optimal fitting model sets meeting the error threshold in querying time. Combining the above tow algorithms, this paper finally proposes an efficient and confidence query processing algorithm based on spatial-temporal correlations, theoretical and experimental results show that this method can return high confident querying results and is energy efficient.

Key words: wireless sensor network, spatio-temporal, time slice fitting, confidence and efficient query