In the BONSAI Laboratory there are research and testing activities related to three main strands:
Broadband Optical Switching
- The lab studies various issues related to optical networks and optical switching. In particular, the activity on optical networking covers the following topics: modeling, network design and simulation, system interoperability, optical transparency, control plane, synchronization and time distribution. The activity on switching covers the following topics: optical switching architectures, optical Giga-Ethernet synchronous-switching systems for core and metro applications, optical-switch synchronization subsystems, optical backplane and optical interconnections.
- Lab facilities supporting the networking activity include personal computers mainly dedicated to simulation activities.
- The optical-switching activity is supported by a hardware testbed equipped by prototypes of optoelectronic switches and synchronous IP routers.
- We generate and measure traffic flows through the real applications (for example p2p streaming video, web browsing, attacks) that we want to be able to recognize.
- We train machine-learning real-time recognition algorithms.
- We evaluate the performance of recognition algorithms in terms of true and false positive rates and other relevant metrics. We enhance the existing algorithms by adding new features.
- We design and implement system architectures for the real-time classification of flows.
Transport with Quality of Service of multimedia contents
- We study the transport of multimedia content through the Internet, both wired and wireless. We propose and analyze new methods for resource planning, admission control, traffic policing, shaping and scheduling to guarantee end-to-end QoS.
- We study complex multimedia traffic such as MPEG-coded video, which exhibits long-range dependence and self-similarity, particularly difficult to account for while provisioning resources for guaranteeing the end-to-end QoS.
- We analyze p2p video streaming systems with the objective of defining new rewarding mechanisms for users with powerful CPU resources and high-capacity Internet connections, in such a way to provide better throughput and end-to-end QoS to video streams.