The addition of limited, but nevertheless real, communication capabilities of smart devices, and the increasing deployment of sensors have led to the emergence of the Internet of Things. This field is still an interesting research subject today, because many problems remain to be solved, in particular concerning the interoperability of proposed solutions. The LRT team has assets to propose approaches because the very foundations of IoT mobilize both network and software skills. The work of the LRT team, and more specifically those of Sylvain Cherrier, Etienne Duris and Rémi Forax are moving towards a convergence between service approaches (choreography or orchestration) and knowledge in terms of virtual machines.
An IoT platform has been set up, allowing experiments using collaborations within a collection of virtual machines, voluntarily limited in expressivity, but adapted to the needs of applications related to IoT as well as to strong ones. Constraints of objects involved, in terms of energy and processing power are taken into account in this approach. The platform also offers opportunities for interactions with other themes of the LRT team (mobile networks in particular), but also collaborations within the LIGM laboratory (the different objects are programmed using finite state machines).
This platform promoted the team's participation in two ITEA2 projects (WoO: Web of Objects and SITAC: Social Internet of Things for Apps by the Crowd), as well as in the FUI SCORPION project (in progress). It has also been the subject of a maturing contract with SATT Idf-Innov, concerning the financing of 2 research engineering positions over a period of 9 months.
Among the results obtained, we can mention:
- The description of the model used by the demonstration platform, its effects with regard to the constraints of the networks and IoT devices
- The contributions of transducers'usage and the proposed evolutions to answer the limits of the choreographies of services compared to the orchestration approach
- Analysis of possible inconsistencies in services choreography, especially if they are used on unreliable network infrastructures (such as those encountered in IoT) and the proposal for an overlay of verification and to stabilize the logical state of the participating elements.
As part of his activity concerning the Java platform, the work of Rémi Forax (an expert in this language) focused on the solutions put in place to solve the problems raised in the Java Specification Request (JSR). He works in particular on the JSR 292 dealing with the management of dynamic types. The results presented concerning the implementation of this JSR, then those about optimizations of allocations, have been integrated into the official platform of Java. Rémi Forax's activity and his involvement in the source code of the virtual Java market are important and the visibility of this activity is significant.
Lastly, Etienne Duris' work on redundancies in computer codes makes it possible to obtain metrics on repetitions and a qualification of the similarities between the different functions. This work will be useful in the search for optimization of software quality, especially in constrained environments, which is an item that the team will focus on in the future.