Section: New Results

Communications with vehicles

Vehicular Communication Architectures Design

Participant : Thierry Ernst.

The purpose of this research is to investigate the use of IPv6 protocols for ITS communication architectures. IPv6 is assumed as the de facto version of the Internet Protocol where Internet-based communications are necessary in vehicular networks. This is motivated by the need from the ITS industry for an extended address space and the enhanced mechanisms that only IPv6 provides. The use of IP (whatever its version) is still questioned by the automotive industry for time-critical safety applications. A performance evaluation of IPv6-based communication systems must thus be conducted to study IP applicability for packet flows with strong QoS and security constraints. This year we have significantly pushed for integration of IPv6 features and we contributed to ITS communication architecture standards specification, particularly at ISO TC204 WG16 [53] and COMeSafety (European Specific Support Action project proposing a reference ITS communication architecture for European projects) [52] . This work is partly done in the framework of the CVIS European project which can be described as a proof of concept of ISO CALM. As such we evaluated the system. in [27] .

Cross-Layer Architectures in Mobile Environment

Participants : Olivier Mehani, Thierry Ernst.

Communication systems for mobile devices have reused a lot of the concepts and infrastructures which were developed for “legacy” wired network. Though functional, this approach is far from optimal and there is a need for network systems able to cope with, or even leverage, the inherent mobility of their hosts. Numerous designs of the network stack of protocols, sharing more information than what the OSI model suggested, have been proposed. These approaches are usually referred as cross-layer design . Early 2009 work have been focusing on the development of such a system, allowing to adapt a transport protocol to temporary disruptions due to mobility [38] . The limited applicability of cross-layer designs and their possibly degrading impact on the performance of the rest of the system however makes them hard to implement in complete systems. Additionally, network information needed for proper adaptation may not be available at a single node, as mobile devices could be parts of a mobile network (NEMO, e.g. in a vehicle) and not necessarily direct witnesses of network access changes. This calls for a distributed architecture akin to a cognitive network to better share information and foster better adaptation to the current conditions. An initial proposal of such an architecture was presented in [39] . Future work covers both areas of decentralized context information exchange within the NEMO and that of local decision based on this information. This work is related to the design of cross-layer architectures for vehicular networks at the standardization level (ISO CALM and ETSI TC ITS).

Geographic routing and addressing

Participants : Thierry Ernst, Yacine Khaled, JinHyeock Choi, Ines Ben Jemaa, Manabu Tsukada, Jong-Hyouk Lee, Satoru Noguchi, Hongliang Zhang.

This topic has been launched in October 2007 and is driving an important part of IMARA's resources given our commitments in the GeoNet project. The purpose is to study the combination of IPv6 and geographic-based routing into a performant communication architecture for vehicular networks. Since packets are delivered to multiple nodes, we are relying on multicast addressing and routing at the IP layer. However, IP multicast currently lacks the ability to route packets according to geographic information. We have therefore proposed several approaches to encode geographic information into IP multicast and we published [33] , [32] on this topic. In addition to this, we are analyzing the feasibility and the performance of IP multicast when deployed in vehicular networks. The study of the deployment of IP multicast in VANETs is reported in [57] .

Security and Location Privacy in vehicular networks

Participants : Jong-Hyouk Lee, Thierry Ernst, Olivier Mehani.

In the context of the GeoNet project, we developed a mechanism (Mobile Network Prefix Provisioning, MNPP in short) to advertise the IP address space of a vehicle to adjacent vehicles. This allows establishing a direct communication path between the vehicles. MNPP has been documented as an IETF Internet Draft [55] . However, exposing the address space of the vehicle over the air causes a number of security threats and location privacy issues that we are studying in the context of the MoboSeND project. We are thus developing a security mechanism to perform the advertisement of the address space in a safe way, with the constraint of infrastructure-less networks were nearby nodes don't know each other at first. We also conducted other work also focused on security and trust establishment for routing protocols within a vehicular ad-hoc network [22] .

Combination of NEMO and MANET

Participants : Manabu Tsukada, Yacine Khaled, Thierry Ernst.

Mobile Adhoc Network (MANET) routing protocols and network mobility (NEMO) support protocols are used in vehicular communications; MANET for vehicle-vehicle communications, and NEMO to maintain the Internet access for vehicle-infrastructure communications. The necessary interaction between MANET and NEMO (MANEMO) brings a number of benefits in terms of improved routing (routing optimization) and improved network accessibility (multihoming). However, protocols have been specified independently from one another and their interaction brings a number of technical and scientific issues. Areas of investigation include:

• Selection of the appropriate path when multiple access technologies are available (multihoming) and when both multi-hop vehicle-vehicle or direct vehicle-infrastructure communications are possible;

• Routing optimization: mobility management usually requires routing through some mobility support server in the Internet, which could lead to routing inefficiencies;

• Network mobility has the particularity of allowing recursive mobility, i.e. where a mobile node is attached to another mobile node (e.g. a PDA is attached to the in-vehicle IP network). This is referred to as nested mobility and brings a number of research issues in terms of routing efficiency.

These topics are currently studied as part of a doctoral thesis which started in September 2007. We developed a tool to evaluate the performance [17] , [45] . Our current results have been published in a number of papers: [13] , [32] . Our IETF standardization activities [56] , [51] are also related to this topic.

A messaging system for Peer-to-Peer applications in vehicle communication systems

Participants : Satoru Noguchi, Thierry Ernst.

In this research work we study (1) Cross-layered P2P messaging system for applications (2) IPv6 Service discovery; Harmonization between IPv6 (with mobility extensions) and applications. The objective is firstly to provide a common messaging mechanism for all applications in order to avoid an overlap of multiple messaging mechanisms in the application layer and second Regarding (2): It also provide service discovery mechanism for not only application layer services but also for network layer services. Currently we are collecting application requirements. This study is performed under the umbrella of the CVIS and Geonet workshops and as such we are studying now how to integrate our work into CVIS (CALM) stack.