Nowadays mobile equipments are used widely and connected to the Internet ubiquitously. Billions of sensing devices are deployed for collecting context information in order to support intelligent and realtime decision support. Besides functionality specification, most devices have spatial and temporal constraints. In this setting, how to organize sensing devices properly and thus to achieve the discovery of devices efficiently with respect to the requirement of certain applications is a challenge. To remedy this issue, this paper proposes a multiple index method to organize sensing devices considering their functionality specifications, spatial and temporal attributes, and develops a query mechanism leveraging this multiple indexes. Experiment results show that this technique is effective and efficient