|Presentation||People||Projects||Publications||News||Collaborations||Intranet||Spin-off||EIT ICT Labs||Contact us|
Ad hoc networks are systems that can establish communication sessions between users without using any kind of infrastructure. Hence, the network must be self-organized and dynamically adaptive. Nodes in ad hoc networks are in constant movement and delivering packets to a specific destination becomes a complex task. One of Hipercom's major protocols has been standardized and and is currently in wide use globally: OLSR (Optimized Link State Routing protocol). We have also developed many topics related to ad hoc networks: Quality of Service, Security, TCP over multi-hop, energy, and vehicular networks.
Quality of service involves finding routes between two nodes in the network that satisfies a number of constraints. These constraints could be the requested bandwidth, the maximum delay, the minimum loss probability, the reliability of links, etc. This problem is NP-Complete because it combines additive metrics in the optimization problem. Hipercom proposed heuristics for finding routes that respect up to four metrics when calculating routes between source and destination. Another QoS issue is the creation of models that estimate the actual value of a metric. For example, computing the available bandwidth or the transfer delay on a link, etc. is very complex in a non-deterministic medium access such as Wi-Fi. To resolve this problem, we developed a model for estimating the available bandwidth in a wireless network. This model is based on considering interfering links in cliques, after which we provide the maximum capacity that could be deployed in a clique. We may still enhance the model by adding a scaling factor to the clique equations in order to become more accurate when compared to real measurements.
In ad hoc networks, security is a very important issue since routing nodes are anonymous. In this case, any node, could change its correct information, insert false information, take the identity of other nodes, etc. All the attacks are very easy because anybody could enter and exit the network and also the medium is wireless and open. Moreover, for the survival of a network, we need the willingness of the nodes in order to route packets to the final destination. If nodes do not cooperate correctly, the routing becomes inefficient. Our solution was to develop two different approaches, one based on intrusion detection that checks the incoherence in the routing protocols and then sends alerts to nodes in order to deactivate the intruders and the second is based on flow conservation that permits to check nodes that avoid forwarding. We introduced the latter property into QoS mechanisms, in order to introduce security as a metric in the routing protocol and to find reliable and secure links.
Transport Control Protocol is an old protocol that ensures end-to-end forwarding between source and destination. With multi-hop networks, TCP suffers from multiple problems such as delays introduced by the routing on wireless simple nodes. From the scalability point of view, changing TCP implies the modification of a billion of TCP/IP stacks that is not possible today. We have been exploring how to find the correct routing protocols in order to optimize the timers' calculation in TCP and increase bandwidth and fairness.
Energy is one of the major topics of next generation networks. The scalability of sensor and ad hoc networks implies that the consumption increase more and more and also changing the devices' batteries becomes impossible to do. The energy optimization is crucial. Our work consist of first considering the energy like a metric in the routing protocol and second by a way of cross-layer communication optimizing the medium access and contention resolution in the MAC layer. Those optimizations should increase the network lifetime in a global way and not only a simple optimization of the node energy.
Vehicular ad hoc networks (VANET) are based on short- to medium-range transmission systems that support both vehicle-to-vehicle and vehicle-to-roadside communications. Vehicular networks will enable vehicular safety applications (safety warnings) as well as non-safety applications (real-time traffic information, routing support, mobile entertainment, and many others). We are interested in developing an efficient routing protocol that takes advantage of the fixed network infrastructure deployed along the roads. We are also studying MAC layer issues in order to provide more priority for security messages which have stringent delivery constraints.