Probing Queries in Wireless Sensor Networks

Approximate Queries with Adaptive Processing

“…In the second case, the system should modify the query in order to provide a nonempty result with accuracy guarantees. This behavior can be obtained, for example, by either changing selection conditions, similarly to what has been done in [68,89] for stored data, or by relaxing the query using a skyline-based approach, similarly to what has been proposed in [72] for stored data and in [95] for sensor networks. As will be discussed further in Section 10.3.2, the basic idea of a skyline-based approach to query relaxation of selection and join operations is to use a relaxing function (usually, a numeric function) to quantify the distance of each tuple (pair of tuples) from the specified condition.…”

“…Sensor networks are a special case of distributed stream management system where each data stream refers to data related to some environmental property collected through the usage of sensors. In such environments, an approach similar to that presented in [72] for relational data has been provided with the aim of solving the empty answer problem [95] (probing queries). The proposed solutions for processing probing queries take into account sensor network features and try to reduce communication costs and energy consumption during the execution.…”

Efficient schemes for similarity-aware refinement of aggregation queries

“…From the five standard SQL aggregate operators, users can specify an aggregate operator along with an aggregate constraint <Z>. Data-oriented high low [13,27,31] Value-oriented mid high [25,33,34] Example 2 Based on Example 2, the scientist might use the following query to find the desired sky region: Clearly, one can see that her aggregate operator and constraint are COUNT( * ) and 1000 objects, respectively. Also, she prefers a result that satisfies the aggregation and similarity constraints equally (i.e., α = 0.5).…”

“…That is, refining predicate a i ≤ x I i into a i ≤ x R i counts as one modification operation regardless of the value x R i and the amount of refinement |x I i − x R i |. In the data-oriented measures (e.g., [13,27,31]), the distance between I and R is based on the data points (i.e., tuples) that are included in the result of each query. For instance, to measure the distance between I and an expanded R, [13] computes the distance between I and all the points in R − I (i.e., the extra points added due to expansion).…”

Pattern mining for query answering in marine sensor data

“…However, the user expects at least a reasonable answer from the query. In that case, the query processing system can recommend users relax the query [12], [13], [14]. For example, the user can be recommended to reformulate the query Q1 toQ1.…”