The use of sensor based applications is in expansion in many contexts. Sensors are involved at several scales ranging from the individual (e.g. personal monitoring, smart homes) to regional and even world wide contexts (i.e. logistics, natural resource monitoring and forecast). Easy and efficient management of data streams produced by a large number of heterogeneous sensors is a key issue to support such applications. Numerous solutions for query processing on data streams have been proposed by the scientific community. Several query processors have been implemented and offer heterogeneous querying capabilities and semantics. Our work is a contribution on the formalization of queries on data streams in general, and on sensor data in particular. This paper proposes the Astral algebra; defining operators on temporal relations and streams which allow the expression of a large variety of queries. This proposal extends several aspects of existing results: it presents precise formal definitions of operators which are (or may be) semantically ambiguous and it demonstrates several properties of such operators. Such properties are an important result for query optimization as they are helpful in query rewriting and operator sharing. This formalization deepens the understanding of the queries and facilitates the comparison of the semantics implemented by existing systems. This is an essential step in building mediation solutions involving heterogeneous data stream processing systems. Cross system data exchange and application coupling would be facilitated. This paper discusses existing proposals, presents the Astral algebra, several properties of the operators.