Section: Scientific Foundations
Scientific Foundations
The POPS research group investigates solutions to enhance programmability, adaptability and reachability of small objects designated as “POPS” (Portable Objects Proved to be Safe). The POPS set includes small devices like smart cards, RFID tags or personal digital assistant which are characterized by limited resources, high mobility and high security level in spite of untrusted environment. The development of applications integrating POPS suffers from lack of “reachability ” of these platforms. Indeed, most POPS are not easy to program and high level of expertise is needed to produce software for such limited operating systems and devices. Moreover, POPS mobility induces sudden and frequent disconnections, long round trip times, high bit error rates and small bandwidth.
We believe that the “system and networking” approach makes sense in the context of portable objects proved to be safe. We have demonstrated that the application-driven approach can be very efficient for customization of embedded software. We propose to focus on the following objectives where sensors are privileged platforms:
Customization of evolving and communicating systems
We propose to address the two following problems based on our work on customization of Java-stems:
- 1.1 Dynamical customization:
It corresponds to what we proposed in the previous proposal: “The ability to adapt a POPS system while it is running (after it has been issued) can be an important differentiation factor to ensure the durability of the system, the update of the applications it can run, and its capacity to adapt to new usages and environments. The more POPS are complex (and expensive) and general purpose, the more these requirements are important. Dynamic configuration is clearly in the objectives of the Camille action since the system is designed to support dynamic extensions of the operating system kernel via techniques such as dynamic linking, on-the-fly compilation, and verifiable typed intermediate language.” In means that we have to study how we can design a specialized system – where functionalities could be removed – which is able to evolve in order to get new functionalities while preserving safety of already installed applications.
Our goal is also to promote the JITS platform for POPS.
- 1.2 Optimization of the communication stack:
The goal is to study the architecture of the communication stack in sophisticated cases (i.e. not limited to serial link) and in particular for objects with wireless capabilities. This is illustrated by the Figures 2 and 5 . In the latter, we illustrate the case of a sensor network application. The “selection engine” takes a “model of the application” (including model of the network) and the “specification of the hardware”. A set of protocols of each layer are evaluated according to “cost function” and the “selection engine” generates the combination of protocols which is optimal for the application and the hardware.
From our interaction with industrial partners, we see that it is also important to consider application layers and in particular http.
Realistic wireless networking
We show that the unit-disk graph model is not realistic and that an excellent and sophisticated algorithm can be jeopardized in the real world. Our approach is to consider realistic physical layer (e.g. Log-normal shadowing model). Moreover, it is necessary to validate results analytically, with simulators, but also with real experimentations.
We propose to investigate the two following objectives:
- 2.1 Position-based algorithms:
The main advantage of this family of protocol is that they are both localized and memoryless. It means that these protocols are robust since they do not need a huge quantity of information (1-hop or 2-hop knowledge in most of cases) and that it can support an arbitrary number of simultaneous flows since intermediate nodes do not need to store information. We will investigate (i) protocols based on geographical coordinates (e.g. GPS coordinates) and (ii) protocols based on virtual coordinates when geographical coordinates. Our goal is to propose energy-efficient protocols with guaranteed delivery for the different kind of protocols: unicast, multicast, data collection (for sensors), topology control, etc.
In particular, we are interested in data collection with data fusion and we think that such techniques can be applied in RFID applications where RFID readers can be assimilated to sensors. Depending on the application, the network can apply different filters in order to limit the amount of data which is sent to the sink.
- 2.2 Hardware-software optimizations:
We will focus on low consumption radio interface for wireless sensors. More precisely, we believe that energy constrained objects can take advantage of smart antenna technologies. This area seems very promising on paper but it raises a lot of implementation problems. In the context of a partnership with IEMN (Institut d'Electronique, de Microélectronique et de Nanotechnologie), we want to experiment the implementation of a full communication stack dedicated to smart antennas (physical layer, MAC layer, etc.).
