Section: Software

SPACES

Participants : Russel Nzekwa, Daniel Romero [ correspondant ] , Romain Rouvoy, Lionel Seinturier.

SPACES is a context mediation middleware that follows the REpresentational State Transfer (REST) principles  [71] . The current implementation of SPACES is based on COSMOS context framework  [64] , [93] and the Comanche web server  [61] . Both, COSMOS and Comanche are based on the Fractal component model and use the Julia implementation of the Fractal runtime environment  [61] .

The main features of the current SPACES implementation are presented below:

1. Ubiquitous connectors : SPACES define connectors that encapsulate the distribution concern. These connectors expose the COSMOS context nodes as REST resources with logical URLs associated, and enable interactions between consumers and producers via different communication protocols and the discovery of the available context sources. The current SPACES implementation supports interaction using the HTTP and twitter  [81] protocols. For the discovery, the implementation uses the Service Location Protocol (SLP)  [75] .

2. Context Representation : Following the REST principles, SPACES support multiple representations of the context information: JSON  [66] , XML and Java serialization.

3. Quality of context (QoC) information : The QoC properties are incorporated as service attributes in the SLP advertisements of the context information.

4. Context selection : The restrictions in terms of QoC of the required context information are expressed as LDAP filters  [96] . SPACES benefits from the LDAP based queries of SLP to select the context providers.

We use the XStream 1.3.13  [60] and JSON-lib 2.2.34  [59] to serialize context information as XML and JSON documents. For SLP and twitter we employ jSLP 1.0.0  [92] and twitter-4j 2.0.6  [102] .