Section: Scientific Foundations

Cross-layer approaches

Many teams over the world are dealing with either one or another of the previously mentioned research domains. However, putting together all skills to research some new concepts at the interface between these domains is less common. The bottom-up strategy will promote the development of realistic models from the node to the network behavior in terms of energy, complexity or performance evaluation. This strategy will also contribute to evaluate the impact of new node designs at the network level. The top-down strategy allows to define realistic scenarios, to develop large scale system requirements and specifications. This strategy will also promote the development of realistic interference models. Other aspects are truly transverse. For instance, ensuring security in distributed networks needs to be designed at the different levels, from the PHY layer security paradigm up to distributed mechanisms over large scale networks.

Our cross-layer framework is driven from three complementary points of view:

Performance and optimization – Performance evaluation and global optimization define a cross-layer axis of our project. In this action, we will be able to merge our contributions on smart wireless networks modeling using combinatorial and stochastic modeling tools. Global optimization is meant to describe system-wide behaviors and provide theoretical bounds on its performance, both by benchmarking the existing solutions and by guiding their improvement which will foster new developments. Our global optimization framework will progressively account for the software radio capabilities of the radio nodes, the properties of resource sharing algorithms and new self-* protocols. Realistic models of the wireless medium will be included, as well as refined models of adaptive protocols. This action will lead to three results : realistic models of smart wireless networks properties, global optimization and performance bounds derivation as well as distributed sub-optimal but feasible algorithms. This cross-layer axis on optimization is a necessity for developing new approaches and tools that are both efficient, provably reliable and relevant to the inherent cross-layer, dynamic and statistical nature of the systems under study.

Security – Security is one of the main cross-layer challenges of the SWING project. Security must be envisioned at each level, from hardware to routing protocols, in order to guarantee an end-to-end comprehensive security strategy. Moreover, in the context of embedded architectures, security related processing must be maintained to the least acceptable energy cost. The main challenges will then be the design of new energy efficient cryptographic primitives (in hardware and in software), the design of security mechanisms for routing protocols in order to preserve the networks from some specific attacks. The band deregulation and the on-the-fly adaptation reduces dangerously the access security. If cooperative mechanisms have to be used, the security of the various applications must be simultaneously guaranteed. Thus, security must be considered from a cross-layer perspective to allow cooperation at the physical layer while still protecting from malicious data access.

Prototyping – In SWING, we aim at addressing the challenges of smart wireless networks not only from a theoretical point of view, but also from a practical one, using simulations and prototypes. From our past experience, we acquired and developed several simulation tools. The CITI laboratory is also equipped with up-to-date radio design platforms allowing to test the embedded software radio systems, evaluate MIMO communications and perform real radio channel measurements. These skills have been acquired thanks to strong partnerships with the industrial community, which we plan to expand via new cooperations with Orange Labs, Alcatel-Lucent and other partners.