You are here

Past Research Areas

2004-2005: Packet Trace Generation

We study methods for packet trace classification. The ultimate goal of this research is to obtain an automatic packet trace generator which could be parametrized for obtaining a synthetic trace that mimics the characteristics of Internet links with different traffic characteristics.
A byside result of our research is a methodology for lossless and lossy packet trace compressions.

2003-2005: Fast Packet Buffers

The main objective is the design of next generation of packets buffers that require a storage capacity for several Gb of data and a bandwidth of several hundreds of Gb/s. Usually, these packet buffers will support Virtual Output Queuing (VOQ), which means that they must manage internal data structures of almost one thousand of queues (one queue for each port and service class).
Our designs are based on hybrid SRAM/DRAM design because they are the fastest packet buffers with worst-case bandwidth found in the literature.

2003-2005: Routing and addressing in ad-hoc

During the past few years the research efforts in the area of communications have been increased. And more recently an important part of these efforts have been fo-cused on the Mobile Ad-hoc Networking (MANET). A MANET is mainly a wire-less network without infrastructure, in which the nodes must behave simultaneously as a host and as a router.

One of the main research issues has been the routing. Routing in MANETs can be classified in topology based routing, and Position based routing approaches. The latter determines relative or absolute positions (e.g. GPS) of the nodes in the network. On the other hand, the former approach determines the routes according to topological in-formation to achieve connectivity. Two of the main protocols obtained in the routing area are the well-known AODV (reactive) and the OLSR (proactive) topology based protocols.

One of the main drawbacks of the topological based approaches is the scalability in reaction to the number of nodes (limiting factor). In MANET every node in the net-work requires to participate as a router and the topological protocols rely on flooding of the control packets through the entire network to create and maintain routes. If the number of nodes grows largely, the number of control packets will increase drasti-cally, increasing the overhead in the network, consuming the scarce bandwidth in the network, and therefore reducing the throughput. Without taking into account the mo-bility of the nodes, this scenario is something similar to what happened to the ARPA-net at the beginning.

We are currently working on solutions to improve the scalibility of the routing protocols in MANETs . One of our proposals is to use a structure like the used in Internet, in which the nodes are aggregated into subnets to be handled as a single entity for routing purposes. Unfortunately, the difficulty to apply this sub-netting structure in MANETs is high, due to their dynamic and distributed nature. However, there are scenarios in MANETS in which the nodes can be grouped follow-ing physical or environmental constraints to apply the aforementioned structure. These formed groups can be considered subnets of a MANET, giving the chance of repre-senting multiple routes of a large number of nodes by a single route. Since the routing information is cutting down, a reduction in the overhead may be obtained. We have demonstrated that using analytical models.

The main challenges in which are working for applying this subnetting structure are: the address allocation under mobility scenarios, the dynamic creation and removal of subnets, and finally the maintenance of the already established sessions when a node moves from one subnet to other (mobility between subnets).

2003-2005: Routing and Internet topology

The topology of any network is a representation of the interconnection between nodes in that network. A node could be a router, an IP-interface or an Autonomous System (AS). An interconnection could be a physical link or a business relationship between Autonomous Systems (ASes). A topology generator is a tool that based on some input conditions, in a synthetic way reproduces this interconnection between nodes. This work addresses the problem of the Internet Topology, specifically the interconnection of the ASes.

1988-2003: ATM networks

The group has been working since 1988, and was created by Prof. Olga Casals.The first research areas we covered were related to ATM networks and traffic management, with our participation in the ATM pioneer project RACE 1022 "Technology for ATD". During these first years we were very much interested in analytical modeling of telecommunication networks, and we worked on Matrix Analytic methods, fluid flow models, Benes method, etc. During the RACE project BAF we designed a MAC protocol with QoS guarantees for an APON. We also collaborated in RACE and ACTS projects EXPLOIT, EXPERT and NETPERF. In these projects we worked with one of the first operative ATM platform in Europe.