Research Profiles

Facing the exponential growth of the intradatacenter traffic, the traditional Data-Center Network (DCN) architectures are not capable to stay ahead of the demand in terms of scalability and lowering costs. In this project, we address the routing problem of the intra-datacenter traffic inside CamCube-based server-only DCNs. The latter are composed of servers only and no additional network equipments are employed. Following the SDN paradigm, we firstly propose a new architecture in which the control plane is hosted in an SDN controller. Next, we propose to emulate the whole DCN using the versatile "Mininet" platform. Then, we formulate the path computation problem, considering the Quality-of-Service (QoS) in terms of the requested bandwidth, as multi-objective combinatorial optimization problem. Next, we propose a thoughtful reformulation for the problem which can be solved using the Branch-and-Cut algorithm. Our proposal, named CamCube Routing Protocol (CRP), yields the optimal routing paths for the considered traffic flows that keep a balanced load on the DCN. Based on extensive emulations using "Mininet" and the "ONOS" SDN controller, the obtained results are very good, compared with the shortest-path approach, in terms of packet error rate and latency.

In a Communication Based Train Control System (CBTC), a central zone controller server (ZC) exchanges signaling messages with on-board carborne controllers (CC) inside the trains through a wireless technology. The ZC calculates and sends periodically to each train its Limit of Movement Authority (LMA), i.e. how far the train can proceed. A CC triggers an emergency break (EB) if no message is received within a certain time interval to avoid collision. Clearly, it is not desired to have an EB due to signaling messages losses (called spurious EB) and not to real risks for the trains. Quantifying the rate of spurious EBs and predicting correctly CBTC system performance are hard tasks with important industrial relevance. This work aims at lling this gap using simulation to better predict CBTC system performance and avoid extra provisioning before deployment. A typical CBTC system implementation for metro by Alstom Transport is considered. New ns-3 modules (CBTC protocol, Video trac generator, multi-channel scanning mechanism, 3D antennas patterns) are developed and a piece of existing code is enhanced. The simulation is also used to investigate the dimension of the radio access networks in a realistic environment (specic modems and access point antennas, radio frequencies, train and track models), another aspect also ignored in the previous literature. Last, our approach can be useful to validate some analytical works.