Section: New Results

Vehicular and mobile applications

Participants : Cédric Adjih, Emmanuel Baccelli, Thomas Clausen, Philippe Jacquet, Pascale Minet, Paul Mühlethaler, Yasser Toor.

Executive summary

We have the following vision: in the future mobile internet and static internet will have their core deeply intricated. This means that mobile ad hoc networks will be attached to the core network, form extension and even be part of it. For example in disaster area, a wireless network could replace the destroyed infrastructure and help to the emergency operations.

With this perspective items such as Autoconfiguration, Security are of crucial importance. However there is a potential conflict between a large population of fixed nodes based on ancient protocol and a smaller but more dynamic population based on new protocols. In the integration both population must cooperate in an hybrid protocol.

The difficulty is to build protocols that are as dynamic and efficient as MANET protocols but can support the legacy of the old and heavy internet protocols. The challenge is nevertheless achievable, because the dynamic part of the network needs less frequent updates from the fixed part of the network. Moreover the natural abondance of resource in the fixed part of the network allows it to support the more frequent updates from the mobile part.

OLSR has been found to be the natural best candidate for this challenge since it gathers dynamic and optimization with internet legacy.

Scientific achievements

Military tactical networks

During year 2011, we conducted several expertises about industrial propoals dealing with OLSR use in military tactical networks.

Protocols for vehicular networks

We have achieved numerous studies and design of protocols for vehicular networks and more spefically for V2V (Vehicle-to-Vehicle) network.

First we have studied the channel occupancy induced by the OLSR proactive routing protocol used in a linear Vehicular Ad hoc Network (VANET). Unlike previous studies, which usually use simulations to evaluate the overhead, we have proposed a simple analytical model to carry out this evaluation. Moreover, we did not evaluate the total overhead induced by the routing protocol as is usually proposed, but, for a given node, the channel occupation induced by the routing protocol.

We have studied flooding techniques for safety applications in VANETs. The typical scenario is the diffusion of an alert message after a car crash in a platoon of vehicles. The packet is diffused with the pure flooding, the multipoint relay (MPR) diffusion of OLSR and a geographic aware protocol. For OLSR we have introduced a variant (Robust-MPR) to improve the reliability. Different realistic scenarios were considered and various parameters such as vehicle density, and background traffic load were scrutinized. We have shown that the Robust-MPR and the geographic aware protocol satisfy the requirements of the safety applications while using considerably less overhead than pure flooding.

We have shown that the geographic aware protocols can be improved for the diffusion of an alert message by using opportunistic routing. We have designed OB-VAN (Opportunistic Broadcast for VANets ) a new protocol that uses this idea. One of the novelty of this protocol is the use of an active signalling technique in the acknowledgement procedure to select the best relay taking advantage of the reception pattern of each message. We have studied OB-VAN in a linear VANET and have shown that it outperforms the flooding for the delay and the amount of overhead. However the delivery ratio of OB-VAN may be insufficient for safety applications. This remark has led to the design of R-OB-VAN which is a reliable variant of OB-VAN. With extensive simulations, we have shown that R-OB-VAN maintains a high delivery ratio even in the presence of packet loss due to shadowing.

We have studied the performance of the Aloha scheme in linear VANETs. This analysis assumes a SINR (Signal over Interference plus Noise Ratio) based model. In this model, we have derived the probability of a successful transmission between two vehicles at a distance of R meters. We have also computed the mean throughput according to Shannon's law. In these two models, we have optimized the two quantities directly linked to the achievable network throughput i.e., the mean packet progress and the density of transport.

Finally, we have studied the utilization of opportunistic routing and shown that this technique is also beneficial for point to point traffic. It decreases the delay and increases the throughput compared with shortest path first routing. Moreover, we have also shown that opportunistic routing for point to point traffic eases considerably the optimization of the MAC scheme e.g. the transmission probability for Aloha and the carrier sense threshold for CSMA.

Collaboration

We received support from MoD for this activity.