Section: New Results

Distributed Context Monitoring

In 2014, we obtained some new results in the area of distributed context monitoring solutions to support the development of self-optimising software systems. Context monitoring has emerged as a key capability in various domains to connect software systems to the underlying hardware platform or to the physical world (in the case of ubiquitous systems). In particular, we have investigated to the capability of inferring high-level contextual situations from a large volume of raw data collected from a single device or in the wild. Both hardware (e.g., accelerometer) or software (e.g., performance counters) sensors tend to continuously produce raw data that a context monitoring solution has to quickly filter, process, and convert it into information that can be used by an application or understood by a user.

As a result of the PhD thesis of Adel Noureddine [14] , defended in March 2014, we have developed a middleware toolkit to support in-depth context monitoring in the domain of green computing. In particular, we introduce a software library, named PowerAPI , that can estimate the power consumption in real-time at various granularities of software: from system processes to code methods (see Section  5.3 ). This non-invasive solution provides accurate insights on energy hotspots of software and can be used to derive the energy profile of any software library, thus guiding the developers in optimising the energy consumption of their developments.

As a result of the PhD thesis of Nicolas Haderer [12] , defended in November 2014, we have contributed to the development of a middleware platform to support in-breadth context monitoring in the area of mobile computing. In particular, we promote the distributed middleware solution APISENSE® as an efficient approach to deploy mobile crowd-sensing tasks across a large population of volunteer participants (see Section  5.1 ). In particular, APISENSE® includes a task orchestration algorithm that preserves the privacy and the battery of sensing devices, while maintaining specific sensing coverage objectives (including time and space dimensions). The server-side infrastructure of APISENSE® is generated from a dedicated software product line, while the implementation is based on the FraSCAti platform (see Section  5.2 ).