A Semantics for a Query Language over Sensors, Streams and Relations

Abstract

Abstract. We introduce a query language over sensors, streams and relations and formally describe its semantics. Although the language was specifically designed for sensor network querying, where data is pulled into streams, the semantics contributed in the paper also encompasses the case in which data is pushed onto streams or else lies stored in classical relations. The approach taken is that continuous queries over streams are an extension of classical queries over stored extents. Apart from the fact that query evaluation over streams is reactive, or periodic, the main difference is the conception of windows as an additional collection type with the consequent use of type converter operations to and from streams and windows (which, as bounded collections of tuples, can be operated on in a relational-algebraic setting). The language and the semantics we provide for it advance on previous work in being more comprehensive with respect to the collection types allowed and in being more flexible as to the number and content of the windows contributing to the result at each evaluation event of a continuous query. The formalization advances on previous work in clarifying the implementation onus.

Validated cost models for sensor network queries

“…SNEEql [2] is a declarative query language for WSNs inspired by expressive stream query languages, particularly CQL [1]. Fig.…”

“…1 The material in this section is closely based on previous work [2,6,7] by the authors. 2 It is beyond the scope of this paper to explain how such quality-ofservice expectations are taken into account.…”

“…2 It is beyond the scope of this paper to explain how such quality-ofservice expectations are taken into account. The interested reader is referred to [7] for details.…”

“…The logical algebra underlying SNEEql was given a formal semantics in [2]. In this paper, we firstly show how physical operators (i.e., concrete algorithms for WSN execution) can be assigned to the logical algebraic operators.…”

Comprehensive Optimization of Declarative Sensor Network Queries

“…1); (iv) expressing joins, where the tuples come from different locations in the SN, as a single query and without using materialization points (as would be required in TinyDB); and (v) allowing QoS expectations to be set for the optimizer, such as acquisition rate and delivery time. A detailed description of the SNEEql language, including a formal semantics, is given in [18]. …”