Networks and Distributed Systems: Project Ideas

Networks and Distributed
Systems: Project Ideas
Project Areas
Network security
Privacy in distributed systems
Intermittent connected systems
Traffic control networks
Overlays and CDNs
Wireless networks
Root cause analysis in distributed systems
Network topology evolution
Internet routing
• Convergence
– BGP has exponential convergence time
– Can we design a protocol with better convergence
• Security
– BGP security proposals are hard to deploy
– Defensive policy filters: Can we set routing policy
filters defensively to increase the barrier to attack?
Enterprise routing
• Mismatch between control plane and data plane
– Filtering and routing are treated as different
problems in different planes!
• Class-based routing
– Associate traffic with different classes and perform
configuration on a per class basis
• Decentralized Secure DNS
• Computing trust reputation in decentralized
• Consistent Groups – a framework for integrating
security in different distributed systems and
routing protocols
• Security in Social networks
• Privacy preserving information systems
Traffic signaling networks
• Road traffic in various cities is a big problem!
• Can we inter-network traffic signals within a
city to provide better traffic signaling?
– View it as a networking problem with buffers!
Intermittent Networks
• What if your network is not up all the time or
communication is very intermittent
Cell phone SMS
Power shut down
Lack of wireless signal
• How do you design distributed systems in
intermittent networks?
– Disconnected ATM
– Searching medical records
Overlays and CDNs
• Building an overlay for online gaming
– OverQoS: providing QoS using overlay networks
– Can we provide better guarantees for gaming traffic?
• Intelligent handling of packet loss
• Multicast of packets
• Low bandwidth CDNs
– How do you adapt a CDN to operate when you have
variable bandwidth characteristics on different
• Very low bandwidth links near the end-host
• Layered coding of video
Wireless networks
• Wifi-based Long distance networks
– How do you make WiFi work over 100 kms?
• Use directional antenna
• Change the MAC protocol completely
– How do you build large distribution WiLD
• Urban mesh networks
– How do you deal with interference in urban mesh
• Too many access points
– NYU wireless testbed effort!
Root cause analysis
• Quickly detect where failures occur in
distributed systems
Internet Routing
Financial systems
Server farms
Large code bases
• Using machine learning techniques to build a
statistical toolkit that can be used in different
Network topology evolution
• Internet topology models are very inaccurate
– Do not have a good sense of economics embedded
in them?
• Can we design an economic model that can
reasonably predict Internet topology and traffic