Current Research
- Data Streaming (Professor Xu)
Data streaming is concerned with processing a long stream of data items
in one pass using a small working memory in order to answer a query
regarding the stream.
We are investigating novel paradigms and mechanisms that allow us to
perform large-scale distributed data streaming on tens of thousands
of high-speed links and nodes, and aggregate, compress, and interpret
these streaming results, for better measurement and monitoring of large
networks, which traditional data streaming algorithms do not provide.
- Delay Tolerant Networks (Professors Ammar and Zegura)
There exists many scenarios where the network becomes partitioned because
of physical deterrents. In such scenarios, it will be beneficial if an
intermediate node can buffer and carry a message until it has a chance to
forward them. Our work is concerned with the development of a novel
Message Ferrying (MF) scheme, inspired by its real life analog,
that implements this store, carry and forward routing paradigm. In the MF
scheme, a set of mobile nodes called message ferries take responsibility
for carrying messages between disconnected nodes.
[More]
- Future Internet Architecture (Professor Feamster)
- Intelligent route control (Professor Dovrolis)
Intelligent Route Control (IRC) allows participating networks to avoid
the inefficiencies and hotspots of the current Internet; by using multihoming,
overlay routing, or a combination of both. IRC systems often optimize the cost and performance of
outgoing traffic, based on measurement-driven dynamic path switching
techniques. We are involved in the investigation of route oscillations
caused by such a selfish behavior and in the improvement of the achieved
end-to-end performance.
- Internet measurement (Professors Dovrolis and
Feamster)
- Network estimation (Professor Xu)
Traffic and loss matrices are the most important network performance
statistics that need to be accurately measured and carefully monitored
in an Internet service provider (ISP) network. The accuracy of existing
techniques for estimating traffic matrix, which primarily try to infer
from measureable network statistics such as Cisco NetFlow
records, is not very high.
We propose to develop novel statistical signal processing techniques
that allow us to infer traffic and loss matrices as accurately as possible
from the data we already have in hand.
- Network security and operations (Professor Feamster)
- Diagnosis and troubleshooting
- Network monitoring, security, and defense [SIGCOMM
'06 Paper]
- Fault detection [rcc]
- Overlay networks (Professors Ammar, Dovrolis and Zegura)
Overlay networks help overcome functionality limitations of the Internet
(like quality-of-service, wide-area multicast support), by forming a
virtual network on top of the IP network within which the specialized
overlay nodes collaborate. However,
the onset of multiple overlay applications and the surge of overlay
traffic, can lead to several problems, each which
can lead to a deterioration in the overall system performance and stability.
Moreover, the placement of overlay nodes and links tends to become complicated.
Our group focuses on investigation of the multi-layer interaction, topology
placement, the development of new overlay services and testbeds.
[ NetFinder ]
- Research NOC (Professors Clark and Feamster)
- RNOC: Research Network Operations Center
- Router buffer sizing (Professor Dovrolis)
Packet buffers in router/switch interfaces constitute a central element of
packet networks. The appropriate sizing of these buffers is an important
and open research problem, given constraints on the minimum utilization,
maximum loss rate, and maximum queueing delay. We investigate this problem
and the tradeoff between the loss rate vs the queuing delay. We also address
concerns over whether the buffer size can be smaller than the link's
bandwidth-delay product, as small buffers lead to an increase in the loss
rate and an associated deterioration in the performance of certain application.