MPLS Research Group

 Reference implementation capable of creating user defined topologies is available here.

 The current projects in MPLS development efforts are:

We are extending the architecture of our MPLS switching engine to support QoS using DiffServ paradigm.  In addition, to find the path satisfying the QoS requirements of flow aggregates, and to optimize network operations, signaling support is needed that can establish explicit paths (LSPs) in MPLS networks.  This is also being currently pursued.

We are in the process of  setting up a test network that would include a MPLS label-switched network comprising of our Linux-based implementation of Label-Switched Router (LSR) and MPLS implementations for commercial vendors based on standards, connected by multiple links. This would allow us to study the issues like interoperability between different implementations, performance benchmarking.

2. MPLS Signaling for Traffic Engineering.

We are also extending our signaling implementations to support protocols for traffic engineering in MPLS networks. Specifically, we are integrating RSVP-TE implementation with our MPLS  stack.

3. Linux based MPLS Emulator (LiME) 

MPLS Emulator would be a useful tool for researchers to study MPLS components. It would also serve as test –bed to create complex network topologies and understand the interactions between different components of MPLS network.  Our group is developing a Linux based MPLS emulator, LiME. LiME can also be used as Protocol Development Environment.

A reference implementation of MPLS emulator is available here.

4. Traffic Engineering Algorithms in MPLS Network.

Traffic engineering is a major challenge in traditional IP networks where packets are routed along the shortest path and where the shortest path is calculated using hop count metric.  This may lead to congestion hot spots in the network where some links may remain underutilized.

Traffic engineering makes use of alternative paths and balances the load among all the paths. MPLS is an ideal mechanism for traffic engineering as it can explicitly specify the path from the source to the destination.

In our work on Traffic engineering, we are investigating the problem of finding the optimal distribution of all the traffic demands on the various paths available between source-destination pairs.