PhD Thesis Work Plan

Document Sample
PhD Thesis Work Plan Powered By Docstoc
					                            PhD Thesis Work Plan


The Work Plan is divided in 6 Work Packages:

Work Package 1 – Telefonica Internship
Work Package 2 – Build a GMPLS-TE architecture in OPNET
Work Package 3 – Multicast Traffic Engineering
Work Package 4 – Evaluation Phase
Work Package 5 – Changes to the core
Work Package 6 – Writing the thesis


Work Package 1 – Telefonica Internship

I have the possibility of going to Telefonica’s Research Labs in Barcelona for a 6-
month internship. My possible work there can be divided in three activities:

Activity 1.1: Get data from their IPTV network. This data would then be used as an
input to OPNET simulations. This traffic would be inserted either directly or by
creating a traffic model similar to the real traffic patterns.

Activity 1.2: Study their network infrastructure, core and access. This would help me
in building the topology model input for the simulations.

Activity 1.3: Try to understand Telefonica's ideas for the future, both in terms of
service offering – prediction of new services–, service demands, and of network
evolution. Will the access network evolve to fibre-based technology as Fibre To The
Home or Gigabit Ethernet? Will the core need upgrade to accommodate new service
demands? This would be good to evaluate futuristic scenarios.


The following deliverable is due in the end of this work package:

Deliverable 1 – Description of the inputs to OPNET simulations: IPTV service source
model, demand model and network topology. It might lead to a paper (subject to
contribution novelty).


Contingency plan

If the Telefonica internship plan is frustrated, the inputs to OPNET simulations must
come from other sources.

An investigation on recent literature (mainly papers from IMC2007 [18] and P2P-TV
SIGCOMM 2007 [19], but not exclusively) led me to some useful tools:

   •   Work on Live Streaming Workloads:
         o “A hierarchical characterization of a live streaming media workload”
             [1]
           o “An Analysis of Live Streaming Workloads on the Internet” [2]
   •   Available Topology and Topology Generators:
           o Abilene - Visible Backbone [3] ( topology map of Internet2 back-bone)
           o Inet Topology Generator [4]
           o GT-ITM Topology Generator [5]
           o Tiers topology generator [6]
           o BRITE [7]
   •   Internet traffic / data
           o The Internet Traffic Archive [8]
           o Internet Measurement Data Catalog [9]
           o Predict [10]
   •   Network traffic monitors
           o tcpdump [11] – to intercept packets
           o Bro Intrusion Detection System [12] - passive monitor of network
               traffic

Andrew Moore (Computer Laboratory) is working on Traffic Classification. He has
developed a traffic classification scheme with some of his PhD students that could
serve as the input for application modelling. This might be useful to model the IPTV
service.

The Centre for Photonic Systems is involved in the HIPNet project [13], and this
could also be a good source of information and of joint collaboration, specifically
concerning their work on traffic modelling.

A possible source of data could be UKERNA/JANET [14]. Freewire [15] is offering
60 broadcast-quality TV channels over JANET using multicast so this network could
be a good source of IPTV data.

OPNET will be used for the simulations. However, there are also network emulators
(Modelnet [16]) and testbeds (PlanetLab [17]) available that, if necessary, can also be
considered for this work.


Work Package 2 – Build a GMPLS-TE architecture in OP ET

OPNET will be used to simulate a GMPLS-TE infrastructure. RSVP-TE will be
implemented as the signalling protocol. Since we are interested in optical core
networks, the integration of the optical layer with the multicast algorithm is
fundamental. Each optical fibre will be multiplexed into multiple wavelengths, and
each wavelength can still be sub-multiplexed into several virtual light paths. The
mapping of these light paths with the IP/GMPLS topology will also be taken into
consideration.

The preliminary research work conducted in the first year will be used as foundation
for this work package. This work package can be divided in three activities:

Activity 2.1: Implement the RSVP-TE signalling protocol infrastructure in OPNET.
Activity 2.2: Implement the GMPLS-TE infrastructure using RSVP-TE as signalling
protocol in OPNET.

Activity 2.3: Integration of the multicast TE heuristics in the GMPLS-TE
infrastructure.


This work package will happen concurrently with other work packages because the
infrastructure will be improved as the heuristics (work package 3) are implemented.


The following deliverable is due in the end of this work package:

Deliverable 2 –Description of the GMPLS-TE infrastructure.


Work Package 3 – Multicast Traffic Engineering

In this phase one (or several) heuristics will be implemented in OPNET. Since Traffic
Engineering is typically coarser grained in space and timescale than QoS the
heuristics will be based in QoS multicast protocols (e.g. YAM and QoSMIC).

The heuristics implemented in OPNET will then be validated. This will be done by
simulating a very simple topology and source model and comparing the simulation
results against the numerical (or analytical) expected outputs.

Activity 3.1 – Implementation of multicast TE heuristic in OPNET.

Activity 3.2 – Validation of the simulations.


The following deliverable is due in the end of this work package:

Deliverable 3 – Description of the multicast TE heuristic and validation of the
simulations.


Work Package 4 – Evaluation Phase

The basic evaluation of the heuristic will be done against a) multiple unicast and b)
broadcast solutions. Present scenarios will be verified. Looking at today's Internet
topologies, access technologies, and traffic patterns, one will check how Traffic
Engineering can improve quality for some (or all) users. This stage is used to verify if
simulations produce sensible results.

Then futuristic scenarios will be added: 1) different access technologies - from DSL
to GigE to FTTH, 2) different service demands, with diverse broadcast video quality
(from low - 2 Mbps - to very high quality - 80 Mbps -, for instance), and possibly with
several channels running simultaneously in each home (viewers might view two or
three channels, record several, run video conferences, etc.) In this phase we must
check if multicast TV with Traffic Engineering will support higher quality for more
people. This is the crucial result.

The next phases will be the comparison between the multicast TE heuristic and 1)
PIM-SM (the de facto multicast protocol used by most service operators), 2) QoS
protocols (such as YAM and QoSMIC) and 3) peer-to-peer (and peer-assisted)
algorithms.

All simulations will run for a range of values (and randomised around them) to get
confidence limits on the output, and sensitivity analysis will also be done.

Activity 4.1 – Basic evaluation of the multicast TE heuristic against multiple unicast
and broadcast in present scenarios.

Activity 4.2 – Evaluation of the multicast TE heuristic against multiple unicast and
broadcast in futuristic scenarios.

Activity 4.3 – Evaluation of the multicast TE heuristic against PIM-SM.

Activity 4.4 – Evaluation of the multicast TE heuristic against QoS multicast
protocols.

Activity 4.5 – Evaluation of the multicast TE heuristic against peer-to-peer (and peer-
assisted) algorithms.


The following deliverables are due in the end of each activity:

Deliverable 4 – Basic Evaluation of Multicast with Traffic Engineering

Deliverable 5 – Evaluation of Multicast with Traffic Engineering in futuristic
scenarios.

Deliverable 6 – A comparison between Multicast with Traffic Engineering and PIM-
SM.

Deliverable 7 – A comparison between Multicast with Traffic Engineering and QoS
multicast protocols.

Deliverable 8 – A comparison between Multicast with Traffic Engineering and peer-
to-peer (and peer-assisted) multicast.

Some deliverables can eventually lead to a paper (subject to contribution novelty and
results achieved).

Work Package 5 – Changes to the core

In this work package I will investigate how the core network evolves to cope with
increased demand (futuristic scenario from the previous work package).
Activity 5.1 – Changes to the core

The following deliverable is due in the end of this work package:

Deliverable 9 – Changes to the core to cope with increased IPTV demand.


Work Package 6 – Writing the report

In the end of all simulations and experiments I will wrap everything up and write my
thesis.

Activity 6.1 – Writing the PhD Thesis


To summarise, the timeline of this thesis proposal is presented in the following
GANTT chart.




                        Figure 1. Timeline of PhD Thesis Proposal




LIST OF REFERE CES

[1] E. Veloso, V. Almeida, W. Meira, A. Bestavros, and S. Jin, “A hierarchical
characterization of a live streaming media workload,” in Proc. of ACM IMC, 2002,
pp. 117–130
[2] Kunwadee Sripanidkulchai, Bruce Maggs, and Hui Zhang, “An Analysis of Live
Streaming Workloads on the Internet”, Proceedings of the 4th ACM SIGCOMM
conference on Internet measurement
[3] Abilene: http://pea.grnoc.iu.edu/Abilene/
[4] Inet Topology Generator: http://topology.eecs.umich.edu/inet/
[5]GT-ITM: http://www.cc.gatech.edu/fac/Ellen.Zegura/graphs.html
[6] Tiers topology generator: http://www.geocities.com/mattdoar/
[7] BRITE: http://www.cs.bu.edu/brite/
[8] The Internet Traffic Archive: http://ita.ee.lbl.gov/
[9] Internet Measurement Data Catalog: http://www.datcat.org/
[10] Predict: https://www.predict.org/
[11] tcpdump: http://www.tcpdump.org/
[12] Bro Intrusion Detection System: http://www.bro-ids.org/
[13] HIPNet project: http://www.swan.ac.uk/iat/CurrentResearchProjects/HIPNet/
[14] JANET: http://www.ja.net/
[15] Freewire: http://www.freewiretv.com/
[16] Modelnet: http://modelnet.ucsd.edu/))
[17] PlanetLab: http://www.planet-lab.org/
[18] Internet Measurement Conference 2007: http://www.imconf.net/imc-2007/
[19] Peer-to-Peer Streaming and IP-TV SIGCOMM Workshop (P2P-TV):
http://www.sigcomm.org/sigcomm2007/p2p-tv/

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:319
posted:6/2/2010
language:English
pages:6