A comparison of BitTorrent and konspire2b peer-to-peer content distribution networks with a traditional client-server content distribution network:

Presentation slides

Several new content distribution networks (CDNs) have recently been introduced. Whereas traditional CDNs employ a hierarchical client-server topology, these new networks use an ad hoc peer-to-peer model. We will investigate the relative performance of two such networks, BitTorrent and konspire2b, and compare them with a traditional CDN. BitTorrent uses a swarmed download approach. Files are partitioned into chunks and nodes seeking a particular file download the chunks from peer nodes, as well as simultaneously uploading chunks to other peers. A tit-for-tat rule is employed to enforce fairness. The availability of chunks amongst nodes is monitored by a centralised entity called a tracker. In contrast, konspire2b distributes files in their entirety. Availability of files is tracked in a distributed manner using message flooding. We propose to simulate both these protocols in a realistic modelling environment that closely mimics the real world. Features such as asymmetric connection bandwidth, random demand for files and transitory connection of peers will be modelled. We intend to quantify performance based on throughput, efficiency, and reliability metrics. We also hope to identify deficiencies in these protocols and suggest improvements.

Simulation tool that you plan to use: ns-2

References:
http://bitconjurer.org/BitTorrent/documentation.html
http://konspire.sourceforge.net
http://www.cen.uiuc.edu/~halm/index.html
http://wiki.theory.org/index.php/BitTorrentFAQ
http://www.rediff.com/netguide/2003/jun/19konspire.htm
http://www.globule.org/

Simulating different search techniques in Gnutella:

Presentation slides and final report (PDF files).

Gnutella is a decentralized data-sharing peer-to-peer (P2P) network. A host initiating a search in the network will send its search request to all its neighbours. Upon receiving the search request, the neighbours will forward to request to all their neighbours. This process continues until a specified radius has been reached. This search method poses a heavy burden on the network because the number of network messages grow exponentially. Various studies has suggested alternative approaches for searching in Gnutella architecture. In this project, these techniques will be simulated.

Since I am not familiar with OPNET and ns2 yet, I have not decided whether I would use OPNET, ns or other simulation tools.

References:
[1] Clip2 (n.d.). The Gnutella protocol specification v0.4. Retrieved from http://www9.limewire.com/developer/gnutella_protocol_0.4.pdf
[2] Kalogeraki, V., Gunopulos, D., & Zeinalipour-Yazti, D. (2002, November). A local search mechanism for peer-to-peer networks [Electronic version]. Series-Proceeding-Section-Article, 300-307.(from attachments)
[3] Yang, B., & Garcia-Molina H. (2002, July). Efficient search in peer-to-peer networks. Paper presented at the 22nd International Conference on Distributed Computing Systems. Vienna, Austria. (from attachments)
[4] Lv, Q., et al. (2002, June). Search and replication in unstructured peer-to-peer networks. Paper presented at the 16th International Conference on Supercomputing. New York City, NY. (from attachments)
[5] Ritter, J. (2001, February). Why Gnutella can't scale. No, really. Retrieved from http://www.darkridge.com/~jpr5/doc/gnutella.html

Performance analysis of voice communication in a private 802.11 network:

Presentation slides and final report (PDF files).

VoIP (Voice over Internet Protocol) phone systems have been gaining popularity in medium to large businesses as replacements of existing PBX (Private Board Exchanges). These VoIP systems usually rely on the Ethernet structure readily available in a business facility. The emergence of wireless LAN (local area network) is also gaining immense popularity due to the additional mobility given to the users. The purpose of this project is to simulate a PBX-style VoIP system employed over 802.11b network. Performance factors including packet jitter and delay will be analyzed. The study shall include different types of voice encoding (G.711, G.723, G.729, etc.) as parameters. If time permits, 802.11g will also be looked at for performance comparison against 802.11b. Also, new QoS (quality of service) standards for wireless networks are being drafted. The effectiveness of these draft protocols will be examined.

The topic of my project is similar to one presented in Sprint 2003, but the approach will be quite different. Real VoIP models will be used. A PBX environment with internal and external calls will be simulated. Background data traffic will be modelled to examine the QoS protocols.

References:

P2P networks as content distribution networks:

Presentation slides and final report (PDF files).

I am interested in evaluating the feasibility of using a P2P network as a specialized form of content distribution network (CDN). I intent to extend the "Gnutella Connection Dynamics" project done in Spring 2003 to add queries and file transfers in order to produce a relevant simulation. I'd like to examine the performance characteristics of this type of scheme, in particular the time it takes for the latest version of content to propagate through the network. Prompt delivery of content is important to serveral types of applications that employ CDNs, such as the distribution of virus signature update files. We'll investigate the variation of this distribution time and the related probability of a client having access to the latest version of the content qua time and various network seeding methods. Naturally, we reserve the right to scale down this project should it prove to be too ambitious given the time constraints.

References:
(1) Gnutella - Stable - 0.4, http://rfc-gnutella.sourceforge.net/developer/stable/index.html
(2) Gnutella - Query Routing Protocol, http://rfc-gnutella.sourceforge.net/src/qrp.html
(3) Bergner, Marcus, "Improving Performance of Modern Peer-to-Peer Services", http://www.cs.umu.se/~bergner/thesis/html/thesis.html
(4) Ivkovic, Igor, "Improving Gnutella Protocol: Protocol Analysis and Research Proposals", http://www.swen.uwaterloo.ca/~iivkovic/Gnutella.Limewire2001.pdf
(5) Horowitz, Ellis, "A Look at Peer-to-Peer File Sharing with Gnutella", http://ilab.usc.edu/classes/2002cs597f/1

Simulation of Handoff Procedure based on SIP over Wireless LAN

Presentation slides and final report (PDF files).

Wireless Local Area Network (WLAN) is becoming a hot topic in recent years. An important issue is the handoff management between different WLAN subnets. Many alternative ways to realize mobility management have been proposed based on different layers. In our project, we are interested in simulating a handoff procedure based on Session Initial Protocol (SIP) over Wireless LAN. We will study analytically the SIP-based handoff delay. 1. E. Wedlund, H. Schulzrinne, "Mobility Support using SIP", ACM/IEEE International Conference on Wireless and Multimedia (WOWMOM), Aug. 1999, pp. 76-82.

References:
2. Banerjee N., Basu K., Das S.K., "Hand-off delay analysis in sip-based mobility management in wireless networks", Parallel and Distributed Processing Symposium, 2003. Proceedings. International , April 22-26, 2003.
3. IEEE, "802.11 Wireless LAN Medium Access Control(MAC) and Physical Layer (PHY) specifications," approved 26 June 1997.
4. M. Handley, H. Schulzrinne, E. Schooler, and J. Rosenberg, ”SIP: Session Initiation Protocol”, RFC 2543, Internet Engineering Task Force, March 1999.
5. OPNET documentation: "Wireless LAN Model Description", MD-36-1 to MD-36-16,Release 9.1

Implementation and simulation of Linux virtual server using ns-2:

Presentation slides and final report (PDF files).

Virtual server is built on a cluster of real servers. End users need not to know the architecture of the cluster, they see only a single virtual server. That is to say, virtual server provides end users with a single entry point to a cluster of real servers.

The project is to implement a virtual server framework, plus an old DNS-redirection framework and a combination of DNS-redirection and virtual server framework. After the implementation, we will evaluate and compare their performances, with an emphasis on the performance of the virtual server when using its three different techniques: Network address translation (NAT), IP tunneling(encapsulation), and Direct routing (DR). Furthermore, we will invesitigate different scheduling algorithms used by the virtual server to balance the load.

References:
1. Linux virtual server for scalable network services, Ottawa Linux Symposium, 2000 http://www.linuxvirtualserver.org/ols/lvs.ps.gz
2. Load Balancing The UK National JANET Web Cache Service Using Linux Virtual Servers by Michael Sparks at UK National Janet Web Cache Service, November, 1999. http://wwwcache.ja.net/JanetService/PilotService.html
3. Performance Evaluation of Linux Virtual Server, Patrick O'Rourke, Mike Keefe, April 2001. http://www.linuxvirtualserver.org/performance/lvs.ps.gz
4. LVS-HOWTO Joseph Mack, http://www.linuxvirtualserver.org/Joseph.Mack/HOWTO/index.html
5. T.Brisco, Dns support for load balancing, http://www.ietf.org/rfc/rfc1794.txt April 1995, RFC1794.

TCF-friendly rate control mechanism: implementation and analysis using ns-2:

Presentation slides and final report (PDF files).

New trends in communication, in particular the deployment of multicast and real-time audio/video streaming applications, are likely to increase the percentage of non-TCP traffic in the Internet. These applications rarely perform congestion control in a TCP-friendly manner; they do not share the available bandwidth fairly with applications built on TCP, such as Web browsers, FTP, or e-mail clients. The Internet community strongly fears that the current evolution could lead to congestion collapse and starvation of TCP traffic. For this reason, TCP-friendly protocols are being developed that behave fairly with respect to coexistent TCP flows. In this project, I present the implementation, simulation and evaluation of a protocol that adopts TCP-friendly Rate Control (TFRC) mechanism [5], for an efficient and desirable reaction to network congestion. I have used ns-2 tool support for this project.

References:
[1] S. Floyd and K. Fall, Promoting the Use of End-to-end Congestion Control in the Internet, IEEE/ACM Trans. Net., vol. 7, no. 4, Aug. 1999, pp. 45872.
[2] Widmer, J.; Denda, R.; Mauve, M.; A survey on TCP-friendly congestion control Network, IEEE , Volume:15 Issue:3, May-June 2001 Page(s): 28 -37
[3] Ramakrishnan, K., Floyd, S. and D. Black, "The Addition of Explicit Congestion Notification (ECN) to IP", RFC 3168, September 2001.
[4] Balakrishnan, H., Rahul, H., and S. Seshan, "An Integrated Congestion Management Architecture for Internet Hosts," Proc. ACM SIGCOMM, Cambridge, MA, September 1999.
[5] Handley, M.; Floyd, S.; Padhye, J.; Widmer, J.; TCP Friendly Protocol Specification (TFRC): Protocol Specification; RFC 3448, January 2003.
[6] J. Padhye, J. Kurose, D. Towsley, and R. Koodli, "A model based TCP-friendly rate control protocol," in Proc. International Workshop on Network and Operating System Support for Digital Audio and Video (NOSSDAV), Basking Ridge, NJ, June 1999. 20 http://citeseer.nj.nec.com/padhye99model.html

Analysis and Trace Driven Simulation of H.323 VoIP Traffic:

Presentation slides and final report (PDF files).

Although Voice over IP (VoIP) has been defined for some time, the adoption rate for this form of voice communication has not been high until recently. As consumers adopt broadband connections and proposals such as Multi-protocol Label Switching (MPLS) and Reservation protocol (RSVP) are implemented to increasing quality of service (QoS), VoIP are again gaining popularity. Previous projects have analyzed VoIP performance using Constant Bit Rate (CBR) traffic, to mimic video/audio streaming applications. Instead, this project's aim is to implement a VoIP application using the H.323 protocol in network simulator (ns-2) which includes many of the signalling functions. Furthermore, the model would include include the capability of running trace-drive simulation based on real-traffic traces.

We want to examine the network and VoIP call performance in such network. Some of the performance parameters we wish to examine are delay and delay variance for VoIP calls, number of sustainable VoIP calls, and delay and delay variance for other services. In addition, we have obtained a number of VoIP traffic traces from an overseas ISP. Using the tcpdump and Cisco netflow traces we will try to compare the genuine traffic with the ns-2 exponential-on/exponential-off model and trace-driven simulation. With the results, we will iscuss the simularities and differences between them.

References:
[1] VoIP over Frame Relay with Quality of Service (Fragmentation, Traffic Shaping, LLQ / IP RTP Priority) http://www.cisco.com/en/US/tech/tk652/tk698/technologies_configuration_example09186a0080094af9.shtml
[2] www.logikom.net/traces
[3] Voice over IP for the Cisco AS5300 http://www.cisco.com/en/US/products/sw/losswrel/ps1830/products_feature_guide_chapter09186a008008808d.htm
[4] NetFlow services solutions Guide http://www.cisco.com/univerod/oc/td/doc/cisintwk/intsolns/netflsol/nfwhite.htm#xtocid7l
[5] http://www.tcpdump.org/
[6] Network Research Group (NRG) at Lawrence Berkeley National Laboratory (LBNL) in Berkeley http://ee.lbl.gov/
[7] H.323 Protocol International Telecommunication Union
[8] J. Bellamy, Digital Telephony, 3rd Ed. NY:NY, John Wiley & Sons, 2000
[9] S. Floyd and V. Paxson, "Difficulties in simulating the Internet," IEEE/ACM Transactions on Networking, vol. 9, no. 4, pp. 392 - 403, August 2001.
[10] V. Paxson and S. Floyd, "Wide-area Traffic: The Failure of Poisson Modeling," IEEE/ACM Transactions on Networking, pp.226-244, June 1995.
[11] W. Leland, M. Taqqu, W. Willinger, and D. Wilson, "On the self-similar nature of Ethernet traffic (extended version)," /IEEE/ACM Trans. Networking,/ vol. 2, pp. 1-15, Feb. 1994.
[12] D. Minoli, Delivering voice over IP networks, Chichester, England: John Wiley, 1998.
[13] D. Collins, Carieer Grade Voice over IP, New York, New York: McGraw Hill, 2001.

Analysis of mesh performance for multi-player video games:

Presentation slides and final report (PDF files).

A significant challenge in running networked video games in which more than two players participate is the need to transmit packets in a mesh form from each client to all other clients at a high frame rate. When users have different connection speeds, those with low bandwidth (dial-up) users can significantly degrade the performance of the game if a standard mesh is used and they need to send packets to each of the other users. One simple way to reduce the bandwidth required of a slow user is to send only one copy of the game state to a faster user and have them act as a rebroadcaster, although this increases the overall delay. This project will use ns-2 to analyze the performance of such networks, and determine the maximum amount of data which can be sent for varying numbers of users / connection speeds. This data will be very useful in designing games as the complexity of the information transferred must be balanced against the number of users / bandwidth the game must support.

References:
RFC 793 Transmission Control Protocol http://www.ietf.org/rfc/rfc0793.txt?number=793
RFC 2018 TCP Selective Acknowledgement Options http://www.ietf.org/rfc/rfc2018.txt?number=2018
RFC 2581 TCP Congestion Control http://www.ietf.org/rfc/rfc2581.txt?number=2581
RFC 3449 TCP Performance Implications of Network Path Assymetry http://www.ietf.org/rfc/rfc3449.txt?number=3449
RFC 768 User Datagram Protocol http://www.ietf.org/rfc/rfc0768.txt?number=768

GPRS: Wireless links, multiple base transmitter stations, base station controller, and cell update

Presentation slides and final report (PDF files).

In this project we will add wireless links and multiple Base Transmitter Station (BTS) support to the existing General Packet Radio Service (GPRS) OPNET model. The current model supports one BTS, multiple Mobile Stations (MS) and a wired connection instead of a wireless link in between them. We have to build a Base Station Controller (BSC) which is situated between the BTSs and the Serving GPRS Support Node (SGSN). The BSC routes the traffic to different BTSs and controls the handover (cell update) between them. We are also planning to simulate the cell change between different BTSs.

References:
[1] Gunnar Heine, Holger Sagkob, GPRS Gateway to Third Generation Mobile Networks, Artech House, 2003, ISBN 1-58053-159-8
[2] Jukka Lempiainen, Matti Manninen, Radio Interface System Planning for GSM/GPRS/UMTS, Kluwer Academic Publishers, 2001, ISBN 0-7923-7516-5
[3] Emmanuel Seurre, Patrick Savelli, Pierre-Jean Pietri, GPRS for Mobile Internet, Artech House, 2003, ISBN 1-58053-600-X
[4] Ricky Ng, Ljiljana Trajkovic, "Simulation of General Packet Radio,Service Network", OPNETWORK 2002, Washington, DC, Aug. 2002
[5] Mikael Johansson, "Simulation of Logical Link Layer in GPRS", Simon Fraser University, Burnaby, Spring 2003
[6] Digital cellular telecommunications system (Phase 2+), General Packet Radio Service (GPRS) Service description, Stage 2 (3GPP TS 03.60 version 7.9.0 Release 1998) Download the standard here
[7] Christoffer Andersson, GPRS and 3G Wireless Applications: Professional Developer's Guide, John Wiley & Sons, 2001, ISBN 04714140580 Read book online

Comparison of route optimization and reverse routing for mobile IP over IPv4

Presentation slides and final report (PDF files).

For the past few years, wireless communications has had a profound impact on our society. Many analysts predict that connecting to the Internet with mobile equipments will soon be the dominant access method. However, the current IPv4 protocol does not provide adequate support of portable IP addresses. As a result, the Mobile IP Working Group of the Internet Engineering Task Force (IETF) introduced extensions to optimize datagram routing for IPv4.

In our project, we will complete the Route Optimization extension for Mobile IPv4 as implemented by Hao (Leo) Chen in Spring 2002. We will complete the remaining Route Optimization messages so that the proposed extension can be fully simulated. In addition, we will implement another type of Route Optimization, namely the Reverse Routing for Mobile IPv4 extension, so that the performance of the two proposals can be compared using ns-2.

References:
[1] C. Perkins, IP Mobility Support, RFC 2002, October 1996.
[2] P. Calhoun and C. Perkins, Mobile IP Challenge/Response Extensions, Internet Draft, Mobile IP Working Group, Feb 25, 1999.
[3] H. Chen and L. Trajkovic, Simulation of Route Optimzation in Mobile IP, SFU, Sept 11, 2002.
[4] P. Zhou and O. Yang, "Reverse Routing: An Alternative to MIP and ROMIP Protocols", Proceedings of 1999 IEEE Canadian Conference on Electrical and Computer Engineering, Volume 1, pp. 150-155.
[5] C. Perkins and D. Johnson, Route Optimization in Mobile IP, Internet Draft, Mobile IP Working Group, Sept 6, 2001.

More efficient routing algorithm for ad hoc network

Presentation slides and final report (PDF files).

Network wide broadcasting in Mobile Ad Hoc Networks provides important control and route establishment functionality for a number of unicast and multicast protocols. Nodes in an ad hoc network must cooperate and carry out a distributed routing protocol in order to make multi-hop communications possible. On Demand Routing is one of the most popular routing styles in ad hoc networks. In On Demand Routing, "flooding" is used to find a feasible route from source to destination. The blind flooding can become very inefficient because of redundant, superfluous forwarding. In fact, superfluous flooding increases link overhead and wireless medium congestion. In this project, we implement and modify the mechanism of multipoint relays (MPR) to efficiently flood the broadcast message in the mobile wireless networks. We demonstrate the efficiency of the proposed scheme in the AODV (Ad Hoc On Demand Distance Vector) routing scheme.

References:
1. Yoav Sasson, David Cavin, André Schiper. Probabilistic Broadcast for Flooding in Wireless Mobile Ad hoc Networks. IEEE Wireless Communications and Networking Conference (WCNC) - March 2003
2. Zygmunt J. Haas, Joseph Y. Halpern, and Li Li. Gossip-based ad hoc routing. In IEEE INFOCOM, Jun 2002.
3. Sze-Yao Ni, Yu-Chee Tseng, Yuh-Shyan Chen, and Jang-Ping Sheu. The broadcast storm problem in a mobile ad hoc network. In Proceedings of the Fifth Annual ACM/IEEE International Conference on Mobile Computing and Networking, pages 151162, Aug 1999.
4. T. Clausen, P. Jacquet, A. Laouiti, P. Muhlethaler, a. Qayyum et L. Viennot. Optimized Link State Routing Protocol, IEEE INMIC Pakistan 2001.
5. Charles E. Perkins, Elizabeth M. Belding-Royer, and Samir Das. "Ad Hoc On Demand Distance Vector (AODV) Routing." IETF Internet draft, draft-ietf-manet-aodv-12.txt, November 2002 (Work in Progress).

Simulation and performance study of Ad Hoc routing

Presentation slides and final report (PDF files).

Ad Hoc Networks is an interesting special case of networks. It is an infrastructureless environment that operates on wireless nodes that gets connected only when the nodes are into proximity with each other. This paradigm has gained much interest in recent years due to several inherent applications and benefits. Therefore, The MANET (Mobile ad hoc networks) group within Internet Engineering Task force (IETF), has been formed to work on the aspects of standardizing the protocols and its internet connectivity. The purpose of the simulation work is to study the performance of the ad hoc network under a given routing protocol (such as AODV) and possible enhancements. For example, to study:

1. Connectivity and Dynamic Topology: Since the network continually operates on the assumption that connections are created only when nodes close, also connections are indirectly created when nodes become routers to create multi-hop paths. Therefore, the topology of the network is dynamic. The simulation study can inspect the dynamics of changes in topology of the ad hoc network and discover the influencing factors

2. Scalability of the algorithm: the routing protocol should not depend on a central node of processing for controlling routing. Instead, the algorithm should be distributed to cope with the fact that routing nodes may leave the network. However, because the algorithm imposes extra routing loads on nodes, there is a need to measure that load. Also, applications, relying on ad hoc networks, require increasing the domain of coverage of the ad hoc network and the number of nodes that it can support. For this reason, the simulation study can help answering the following questions:
a. How many nodes the algorithm can support?
b. How (is it possible) to improve it to support more nodes and what are the parameters that influence scalability of the algorithm
c. If possible to compare two algorithms in terms of scalability.

References:
[1] C. Perkins and E. Royer, "Ad-hoc On-Demand Distance Vector Routing" In IEEE Workshop on Mobile Computing Systems and Applications, February 1999.
[2] C. Perkins, E. Royer, and S. Das, Mobile Ad Hoc Networking (MANET), "Ad-hoc On-Demand Distance Vector (AODV) Routing", IETF RFC 3561.
[3] S. Corson and J. Macker, "Routing Protocol Performance Issues and Evaluation Considerations" in IETF RFC 2501.
[4] I. Chakeres and E. Royer, "The Utility of Hello Messages for Determining Link connectivity", The Fifth International Symposium on Wireless Personal Multimedia Communications, Oct. 2002.
[5] M. Günes, U. Sorges, I. Bouazizi, "ARA The Ant-Colony Based Routing Algorithm for MANETs" Int. Workshop on Ad Hoc Networking (IWAHN 2002), Vancouver, British Columbia, Canada, August 18-21, 2002.

Internet infrastructure security: an evaluation

Presentation slides and final report (PDF files).

In recent years, research into Internet security has focused on securing the information, rather than the infrastructure itself. Protocol suites such as IPSec that use encryption and authentication algorithms to protect data from mailicous intermediaries are but one of a broader arsenal. Indeed, IPSec is powerless against attacks that target the very infrastructure the information is carried across. For instance, if a link is "cut" (either physically, or logically) no amount of encryption will allow that packet to be received by the intended recipient.

Attacks on network infrastructure can be divided into four main categories:

One primary weakness of the internet infrastrucure is the implicit trust relationship established between routers both internal and external to a network. This research project will highlight some of the common weaknesses and possible attacks against network infrastructure, and will conclude with a discussion of some of the potential solutions, such as Secure-BGP, Secure-OSPF, and DNS Security.

This project will use OPNET.

References:
1) RFC 2328: OSPF v.2
2) RFC 1772, BGP
3) "Interconnecting ISP Networks", Chris Metz, IEEE Internet Computing, March/April 2001, pages 74-80.
4) "Public-Key Infrastructure for the Secure Border Gateway Protocol (S-BGP)", Seo, Lynn, Kent, IEEE, pages 239-253.
5) "Securing the Internet Routing Infrastructure", Papadimitratos, Haas, IEEE Communications Magazine, Oct. 2002, pages 60-68.
6) "Internet Infrastructure Security: A Taxonomy", Chakrabarti, Manimaran, IEEE Network, Nov/Dec 2002, pages 13-21.