Spring 2011 – Recent Developments in the Internet

During the last two decades the Internet and mobile networks have evolved to be the main communication networks that people use in their everyday life. These communication networks are now essential infrastructures for all of us. World Wide Web is about 20 years old, Facebook attracts hundreds of millions of users on a daily basis, billions of pictures, videos and music files are transferred in the Internet.

In the 2011 spring seminar on internetworking we will broadly look at the most interesting recent developments, current trends and future predictions of the Internet. Possible topics include.

• Cloud computing
• Interactive social networks and services
• Green ICT and energy efficiency in Internet
• Future Internet
• Emerging new business models

Topics:

1. Node ranking in p2p networks - Tutor: Andrei Gurtov
2. Energy efficiency of LTE - Tutor: Matti Siekkinen
3. Characterizing the relationship between burstiness and energy consumption - Tutor: Jukka Nurminen
4. Modeling of user behavior of modern mobile applications - Tutor: Jukka Nurminen
5. Using home routers as a distributed platform -Tutor: Jukka Nurminen
6. Sensor Networking and Identity-locator split - Tutor: Miika Komu
7. Identity-locator split in Cloud Computing - Tutor: Miika Komu
8. Cache synchronization in distributed network elements with minimal bandwidth usage - Tutor: Sumanta Saha
9. How Content-Type affects the optimal chunk size in redundancy elimination programs - Tutor: Sumanta Saha
10. Similarity estimation in a typical content server - Tutor: Sumanta Saha
11. Handling dynamic traffic loads by a hybrid cloud - Tutor: Yrjö Raivio
12. Cost analysis of a hybrid cloud - Tutor: Yrjö Raivio
13. Offloading mobile binaries, user interfaces, content or applications - Tutor: Yrjö Raivio
14. Privacy in DONA-like content centric networks: - Tutor: Dmitriy Kuptsov
15. Secure Network Coding - Tutor: Ming Li
16. Is Binary XML ready for WSN applications? - Tutor: Nie Pin
17. Benefits and challenges of Mobile Cloud Computing - Tutor: Antero Juntunen
18. Green computing business - Tutor: Sakari Luukkainen
19. Authentication in Cloud Services - Tutor: Sanna Suoranta
20. Session management in Cloud Services - Tutor: Sanna Suoranta
21. Mobile device energy efficiency issues in HTML5 - Tutor: Antti P Miettinen
22. Cloud computing technology for increased battery life - Tutor: Antti P Miettinen
23. The Impact of Virtualization on Performance of Cloud Data Centers - Tutor: Zhonghong Ou
24. Does Broadcast-based Address Resolution Protocol (ARP) Work in Data Center Networks? - Tutor: Zhonghong Ou
25. Survey of MANET-based geosocial networks - Tutor: Vilen Looga
26. Analysis of offloading use-cases - Tutor: Vilen Looga
27. Data Center Traffic Measurements - Tutor: Andrey Lukyanenko
28. Client's Puzzles: view of the future - Tutor: Andrey Lukyanenko
29. Fast Wireless (re)connection - Tutor: Andrey Lukyanenko

Topic by Andrei Gurtov

1. Node ranking in p2p networks

Fair sharing of resources is important in future Internet sytems thatcontain users that both consume and provide services. Distributedranking of nodes is a challenging task when nodes are selfish and noglobal view of the system exists. Provide a survey or propose newmechanisms to ensure fair resource distribution in peer-to-peernetworks. Propose criteria to compare existing solutions

References:

1. FairTorrent: bringing fairness to peer-to-peer systems
   http://portal.acm.org/citation.cfm?id=1658955
2. D. Korzun, B. Nechaev, and A. Gurtov, “Cyclic routing: Generalizing lookahead in peer-to-peer networks,” in AICCSA2009: Proc. 7th IEEE/ACS Int’l Conf. on Computer Systems and Applications. IEEE Computer Society, May 2009, pp. 697–704.
3. D. Korzun, A. Gurtov, A Local Equilibrium Model for P2P Resource Ranking, ACM SIGMETRICS Performance Evaluation Review, 37(2): 27-29, September 2009.

Tutor: Andrei Gurtov

Topic by Matti Siekkinen

2. Energy efficiency of LTE

Background: The 3g network interface of currently used commercial 3G phones consume a lot of power. Specifically, it exhibits a rather heavy so called "tail energy" which wastes a lot of energy especially if the workload is sporadic in nature. The first commercial next generation LTE (Long Term Evolution), i.e. UMTS release 8, handsets and networks have emerged last year. The standard promises clearly higher data rates, but it is also promised to improve energy efficiency.

Objective: In this topic, the student should investigate the potential of LTE to improve energy efficiency compared to the currently used 3G networks. The work is done through literature study of the standard and related papers.

References: provided after the topic is assigned.

Tutor: Matti Siekkinen

Topics by Jukka Nurminen

3. Characterizing the relationship between burstiness and energy consumption

Measurements show that (at least in Nokia phones) traffic sent with the same average speed consumes much less energy when it is sent in a bursty fashion. In this topic the task is to study the existing research on how energy consumption correlates with different levels of burstiness. Are there any models which characterize the relationship? What are the appropriate parameters to characterize the burstiness of the traffic? What kind of measurements would be needed to determine the relationship?

References: provided after the topic is assigned.

Tutor: Jukka Nurminen

4. Modeling of user behavior of modern mobile applications

Performing measurements with live networks is often difficult and controlled repeatable measurements are hard. Two solutions are available for this problem. First, using earlier measurements in a simulated fashion. Second, creating a model that captures the essential parts of the user behavior. Such models of e.g. mobility have been used in radio network research . In this topic the idea is to look for models that would be applicable for modern mobile user behavior. For instance, is the page access pattern of a mobile phone web user fundamentally different from the PC user? What kind of access patterns do social software use (e.g. Facebook) in mobile device cause?

References:

1. Amit Jardosh, Elizabeth M. Belding-Royer, Kevin C. Almeroth, and Subhash Suri. 2003. Towards realistic mobility models for mobile ad hoc networks. In Proceedings of the 9th annual international conference on Mobile computing and networking (MobiCom '03). ACM, New York, NY, USA, 217-229.

2. Mah, B.A.; , "An empirical model of HTTP network traffic," INFOCOM '97. Sixteenth Annual Joint Conference of the IEEE Computer and Communications Societies. Proceedings IEEE , vol.2, no., pp.592-600 vol.2, 7-12 Apr 1997

Tutor: Jukka Nurminen

5. Using home routers as a distributed platform

Most homes today are equipped with broadband routers, many of which have reasonable computing and storage capacity. There have been ideas how the routers could act as nanodatacenters and perform useful tasks. For instance, they could contribute to content distribution in a peer-to-peer fashion. An important question arising from this use is how to ensure that this additional operation of the routers does not harm the normal home use. Investigate what kind of solutions have been proposed and experimented with home routers. How do those solutions ensure that the level of service for the regular users remains adequate? How could we find out how much spare capacity a home router has? Is there a way to distill this information into some simple parameters?

References:

1. Vytautas Valancius, Nikolaos Laoutaris, Laurent Massouli\&\#233;, Christophe Diot, and Pablo Rodriguez. 2009. Greening the internet with nano data centers. In Proceedings of the 5th international conference on Emerging networking experiments and technologies (CoNEXT '09). ACM, New York, NY, USA, 37-48.

2. Kelényi, I., Ludányi, A., Nurminen, J.K., "Energy-Efficient Torrent Downloads to Mobile Phones through Memory-Limited Proxies" 8th IEEE Consumer Communications & Networking Conference CCNC’2011, Las Vegas, Nevada, January 2011

Tutor: Jukka Nurminen

Topics by Miika Komu

6. Sensor Networking and Identity-locator split

Background: Constrained RESTful Environments (CORE) working group standardizes a HTTP-based protocol to interact with sensors (temperature sensors, power meters, door locks, etc). The working group still lacks a protocol to secure such interactions.

Task: Analyze if a protocol realizing the so called identity-locator split (Host Identity Protocol or its "diet" version) meets the requirements of the CORE working group in the IETF. Does the self-managed namespace in HIP offer any benefits compared to other solutions? Compare if TLS or dTLS would make a better fit.

References:

1. http://tools.ietf.org/html/draft-moskowitz-hip-rg-dex
2. https://datatracker.ietf.org/wg/core/charter/
3. http://tools.ietf.org/wg/core/

Tutor: Miika Komu

7. Identity-locator split in Cloud Computing

Background: Cloud computing has become popular but has been under scrutiny for its proprietary nature. As a consequence, a number of open source clouds have arose and standardization activities have been started. In the field of standardization, problems related to APIs, naming, security and service mobility are being actively discussed.

Task: Host Identity Protocol (HIP) implements so called identity-locator split. Analyze and discuss if it can be applied to solve the mentioned problems for cloud computing.

References:

1. http://trac.tools.ietf.org/area/app/trac/attachment/wiki/Clouds/Karavettil-et-al-IETF-Cloud-Security-Framework-07Jan2011_v1a.pdf.pdf
2. http://tools.ietf.org/html/draft-khasnabish-cloud-industry-workitems-survey
3. http://tools.ietf.org/html/draft-khasnabish-cloud-sdo-survey
4. http://tools.ietf.org/html/draft-yokota-cloud-service-mobility

Tutor: Miika Komu

Topics by Sumanta Saha

8. Cache synchronization in distributed network elements with minimal bandwidth usage

In many scenarios, cache synchronization in distributed network elements is a necessity.For example, in routers, the route information needs to be disseminated to the neighboursto build up a global network view; in proxy environments, the knowledge of proxy-ed objectcan be used to pick up efficient uplink request route from the clients; and so on. However, thesynchronization of caches require bandwidth consumption, and this has to be as minimal aspossible to not disrupt the actual user plane traffic.

There has already been many studies [1,2] on this for different scenarios such as distributedcomputing, RFID, routers, and virtualization. The student is expected to study those and reporton the best ones with a proposal on how to convert them to be useful for network element cachesync. The report should include a short summary of the feasible methods, study of those fromdifferent angle such as their expected maximum divergence, cache sync time, bandwidth usage,and outdated entry management. A student with the expectation of a higher grade should alsoinclude a study on how to use any of them in a networked scenario where a finite databaseneeds to be synchronized with minimum bandwidth usage.

References:

1. Chris Olston and Jennifer Widom. 2002. Best-effort cache synchronization with source cooperation. In Proceedings of the 2002 ACM SIGMOD international conference on Management of data (SIGMOD '02). ACM, New York, NY, USA, 73-84.

2. Liu Fagui; Jie Yuzhu; Ruan Yongxiong; , "A framework for lightweight data synchronization on mobile RFID devices," Computing, Communication, Control, and Management, 2009. CCCM 2009. ISECS International Colloquium on , vol.4, no., pp.468-473, 8-9 Aug. 2009

Tutor: Sumanta Saha

9. How Content-Type affects the optimal chunk size in redundancy elimination programs

State of the art redundancy elimination programs work at a lower level than IP packets. Theychunk the packet payload in finer pieces based on content dependent signatures to eliminatesimilar content across different media types and files, which is abundant in the Internet [2].There has been recent developments on applying this technique to eliminate redundancy in anetwork level [1], for synchronizing file server contents [3], and so on. Works have also shownthat the size of the chunk to get the highest similarity matching is variable, and one size doesnot fit all [4].

The student is expected to study the impact of content-type over optimal chunk size. There is awealth of resource in the Internet to detect the content type of an HTTP transaction, for example[5], and the student is expected to use a similar software to detect the traffic type, and anothercustom made software (which I can help with) to experiment with different chunk sizes andsimilarity estimation. The outcome of the experiment should a clear result on whether or notcontent-type has any effect on the chunk size, and if it has, some figures and report to back thatup.

References:

1. A. Anand et al., “SmartRE: an architecture for coordinated network-wide redundancy elimination,” in Proc. of SIGCOMM. ACM, 2009, pp. 87–98.

2. A. Anand et al., “Redundancy in network traffic: findings and implications,” in Proc of SIGMETRICS. ACM, 2009, pp. 37–48.

3. N. Bjørner et al., “Content-dependent chunking for differential compression, the local maximumapproach,” Journal of Computer and System Sciences, vol. 76, no. 3-4, pp. 154–203, 2010.

4. K. Tangwongsan et al., “Efficient similarity estimation for systems exploiting data redundancy,” in Proc. of IEEE INFOCOM. IEEE, 2010, pp. 1–9.

5. OpenDPI, available at: http://www.opendpi.org/opendpi.org/index.html, accessed on 09.01.2011

Tutor: Sumanta Saha

10. Similarity estimation in a typical content server

The topic is primarily for those who love hands on tasks. Rather than summarizing andreporting, this topic demands actual measurement tasks with the help of a similarity estimationsoftware. The idea is a simulate a typical content distribution store (such as Nokia music store,iTunes, Microsoft software distributor, rapidshare, etc) and estimate the similarity among thestored contents. From existing research, we already know that there are around 30% similarityin the Internet content, and our target here is to evaluate this statement in the light of a contentprovider. To get an idea about similar measurement works, the student can take a look at [1,2]

References:

1. Gurmeet Singh Manku, Arvind Jain, and Anish Das Sarma. 2007. Detecting near-duplicates for web crawling. In Proceedings of the 16th international conference on World Wide Web (WWW '07). ACM, New York, NY, USA, 141-150.
2. Himabindu Pucha, David G. Andersen, Michael Kaminsky, Exploiting Similarity for Multi-Source Downloads Using File Handprints, USENIX NSDI, 2007

Tutor: Sumanta Saha

Topics by Yrjö Raivio

11. Handling dynamic traffic loads by a hybrid cloud

Summary:
Traffic loads vary a lot in several applications. For example, peak load on ticket sales system or SMS centre can be 10-1000 fold compared to an average load. Cloud technologies can help to manage the peak loads. Idea is that enterprises use private clouds to handle the average loads, while peaks are executed with help of public clouds.

Research target:
Dynamic traffic handling is not a trivial question. You should evaluate the challenges incorporated. What are the requirements for the hybrid infrastructure? How do you manage the dynamic load changes, based on historic or monitoring data or both? When does the hybrid approach make sense?

References:

1. M. Hajjat, X. Sun, Y-W. E. Sung, D. Maltz, S. Rao, K. Sripanidkulchai and M. Tawarmalani: Cloudward Bound: Planning for Benefitial Migration of Enterprise Applications to the Cloud, SIGCOMM 2010.
2. R. Moreno-Vozmediano, R.S. Montero and I.M. Llorente: Elastic Management of Cluster-based Services in the Cloud, ACDC 2009.
3. J. Gabriellson, O. Hubertsson, I. Más and R. Skog: Cloud Computing in telecommunications, Ericsson Review 1/2010.

Tutor: Yrjö Raivio

12. Cost analysis of a hybrid cloud

Summary:
The cost of the cloud depends on servers, infrastructure, power and network costs. In a hybrid architecture public cloud is the 5th cost factor. Enterprises have a difficult to task to decide whether a private, a public or a hybrid cloud solution is the most economical approach.

Research target:
Define a model to evaluate the costs of a private, a public and a hybrid cloud. What is the optimal solution for each product or company? What is the impact of the energy consumption? Where is the breakeven point for each alternative? You may assume that public cloud charging is based on Amazon EC2 pricing model.

References:

1. A. Greenberg, J. Hamilton, D .A. Maltz and P. Patel: The Cost of a Cloud: Research Problems in Data Center Networks, ACM SIGCOMM Computer Communication Review, vol. 39, no. 1, Jan 2009.
2. A. Berl et al: Energy-Efficient Cloud Computing, The Computer Journal, vol. 53, no. 7, 2010.
3. E. Walker, W. Brisken and J. Romney: To Lease or Not to Lease from Storage Clouds, Computer, April 2010 (vol. 43 no. 4) pp. 44-50.

Tutor: Yrjö Raivio

13. Offloading mobile binaries, user interfaces, content or applications

Summary:
The idea of offloading computation from thin or dumb clients to network servers is not novel. However, applying the same idea for mobiles and clouds is pretty new. There are various research projects ongoing on the topic. Basically we can offload the whole mobile binary (image) to cloud or restrict the offloading to only code blocks, user interfaces, applications or content.

Research target:
You have a choice. You may either make a summary of the main research paths on the area, or you may focus on some subarea of the research. Can mobile offloading save energy? What is the cost of computation vs. communication? What is the optimal solution? What are the most vital use cases?

References:

1. X. Zhang et al: Towards an Elastic Application Model for Augmenting Computing Capabilities of Mobile Platforms, Mobilware 2010.
2. K. Kumar and Y.-H. Lu: Cloud Computing for Mobile Users: Can Offloading Computation Save Energy? Computer, April 2010.
3. R. Kemp et al: Cuckoo: a Computation Offloading Framework for Smartphones, MobiCASE 2010.
4. B.-G Chun and P. Maniatis: Augmented Smartphone Applications Through Clone Cloud Execution, HotOS 2009.

Tutor: Yrjö Raivio

Topic by Dmitriy Kuptsov

14. Privacy in DONA-like content centric networks:

In DONA-like networks the content is retrieved by permanent cryptographic identifier independent of its location in the network. Therefore, the questions to answer here are several: (i) how to resolve a human readable name (e.g.,URL) to cryptographic identifier in a privacy preserving way; (ii) how to retrieve the content from DONA-like network by cryptographic identifier in a privacy preserving way; and (iii) will resulting privacy preserving solution make caching harder/impossible in DONA-like network

References:

1. Koponen, T., Chawla, M., Chun, G.B., Ermolinskiy, A., Kim, H.K., Shenker, S., Stoica, I.: "A Data-Oriented (and Beyond) Network Architecture" SIGCOMM CCR 2007
2. Tobias Lauinger "Security & Privacy of Content-Centric Networking" Master Thesis, Technische Uneversitat Darmstadt, September 2010

Tutor: Dmitriy Kuptsov

Topic by Ming Li

15. Secure Network Coding

Different from the standard store-and-forward network paradigm, network coding allows intermediate nodes to mix received packets before forwarding them. Network coding can maximize network throughput and improve robustness in a distributed manner with low complexity. Due to its generality and its vast application potential, it has generated much interest in information and coding theory, multicasting, wireless communication, cryptography, sensor networks, etc. Unfortunately, the mixing of data inherent to network coding exposes network coding systems to a wide range of attacks. These attacks can prevent network coding from being implemented in practice. The student needs to do a survey on security of network coding, including: 1) find the vulnerabilities of network coding and classify them; 2) give the corresponding solutions in literature.

References:

1. Christina Fragouli, Jean-Yves Le Boudec, and Jorg Widmer. 2006. Network coding: an instant primer. SIGCOMM Comput. Commun. Rev. 36, 1 (January 2006), 63-68.
2. Yaping Li; Hongyi Yao; Minghua Chen; Jaggi, S.; Rosen, A.; , "RIPPLE Authentication for Network Coding," INFOCOM, 2010 Proceedings IEEE , vol., no., pp.1-9, 14-19 March 2010.
3. Han, Keesook; Ho, Tracey; Koetter, Ralf; Medard, Muriel; Zhao, Fang; , "On network coding for security," Military Communications Conference, 2007. MILCOM 2007. IEEE , vol., no., pp.1-6, 29-31 Oct. 2007
4. Kawahigashi, Haruko; Terashima, Yoshiaki; , "Security Aspects of the Linear Network Coding," Military Communications Conference, 2007. MILCOM 2007. IEEE , vol., no., pp.1-7, 29-31 Oct. 2007

Tutor: Ming Li

Topic by Nie Pin

16. Is Binary XML ready for WSN applications?

Binary XML refers to any specification which defines the compact representation of XML (Extensible Markup Language) in a binary format. Using a binary XML format generally reduces the verbosity of XML documents and cost of parsing. Binary XML is typically used in applications where standard XML is not an option due to performance limitations. Other advantages may include enabling random access and indexing of XML documents.

Up to now, there are several standard frameworks proposed by different standard organizations. None of them have been widely adopted yet. This work will survey these solutions and evaluate their maturity and performance for Wireless Sensor Network (WSN) applications.

Expectation:
The comparative analysis report will provide comprehensive studies over selected binary XML solutions based on the identified WSN applications. Significant factors and criteria should be defined to provide a generic evaluation framework. The report should include diagrams, charts and tables to illustrate the structure of each solution and its state progress.

Additional information:
No programming is required. In addition to the literature study, the student should try some existing reference implementations of the selected binary XML standards to gain a close look and hands-on experience.

References:

1. Fast infoset: a standard published by ISO/IEC and ITU-T
   https://fi.dev.java.net/standardization.html
2. Efficient XML Interchange (EXI): W3C Standard
   http://www.w3.org/XML/EXI/
3. Wireless Binary XML (WBXML): Open Mobile Alliance (OMA) standard to allow XML documents to be transmitted in a compact manner over mobile networks http://www.openmobilealliance.org/tech/affiliates/LicenseAgreement.asp?DocName=/wap/wap-192-wbxml-20010725-a.pdf
4. Binary MPEG format for XML (BiM): ISO/IEC 23001-1, used by many ETSI standards for Digital TV and Mobile TV.
   http://www.chiariglione.org/mpeg/technologies/mpb-bim/index.htm
5. Extensible Binary Meta Language (EBML) from Matroska
   http://ebml.sourceforge.net/
6. Binary Extensible Markup Language (BXML): Open Geospatial Consortium (OGC) binary XML encoding specification optimized for geo-related data (GML)
   http://portal.opengeospatial.org/files/?artifact_id=13636

Topic by Antero Juntunen

17. Benefits and challenges of Mobile Cloud Computing

Mobile Cloud Computing - utilizing cloud computing in mobile applications and services - offers numerous potential benefits to application developers and end users. For example, offloading computation into the cloud may save battery power of the mobile device while allowing the execution of more computationally demanding applications. On the other hand, the speed, latency, and congestion of mobile networks present challenges to Mobile Cloud Computing. The focus of this paper is on identifying the relevant benefits and challenges of mobile cloud computing and evaluating their impact. The analysis could be purely technical or include the business impact of Mobile Cloud Computing.

References:

1. Weiss, A. Computing in the Clouds. netWorker, 11 (4), pp. 16-25, 2007. doi: 10.1145/1516241.1516350.
2. Perez, S. Why Cloud Computing is the Future of Mobile. ReadWriteWeb.
   http://www.readwriteweb.com/archives/why_cloud_computing_is_the_future_of_mobile.php
3. Mun, K. Mobile Cloud Computing Challenges. Techzine.
   http://www2.alcatel-lucent.com/blogs/techzine/2010/mobile-cloud-computing-challenges/
4. Greengard, S. Cloud computing and developing nations. Communications of the ACM, 53 (5), 2010. doi: 10.1145/1735223.1735232.

Tutor: Antero Juntunen

Topic by Sakari Luukkainen

18. Green computing business

The electricity consumption of ICT service infrastructure is increasing andit is necessary to improve the effectiveness of the energy usage of ICTdevices and infrastructure. Electricity consumption is a significant costitem for the companies in this field and the importance of managing it isenhanced as a competitive advantage.

End user companies have increasingly started to outsource their ICTinfrastructure to large-scale, centralized data centers managed by aspecialized service provider. This trend is further enhanced by theproliferation of cloud computing technologies. Therefore, in recent yearsthe research into electricity consumption of data centers has becomeespecially active. The goal of this research is to study the influence ofenergy effectiveness in the provision of green ICT services and relatedservice providers business models.

References: provided after the topic is assigned.

Tutor: Sakari Luukkainen

Topic by Sanna Suoranta

19. Authentication in Cloud Services

Usually, web service is authenticated using X.509 certificate issuedby some trusted third party. Clients are often authenticated usingpasswods. However, small service providers do not want to pay more forverified service identity than for server capacity. Often cloudservices are said to be easy way to provide new services without heavystarting costs. How do the user and the service authenticate eachother in cloud services? What possibilities different cloud servicesoffer for authentication? Can services be users of other services inthe clouds and how users are then authenticated?

References:

1. Hongwei Li, Yuanshun Dai, Ling Tian and Haomiao Yang Identity-Based Authentication for Cloud Computing CLOUD COMPUTING Lecture Notes in Computer Science, 2009, Volume 5931/2009, 157-166, DOI: 10.1007/978-3-642-10665-1_14
2. S. Subashini and V. Kavitha Review: A survey on security issues in service delivery models of cloud computing Journal of Network and Computer Applications Volume 34 Issue 1, January, 2011 Academic Press Ltd. London, UK, UK DOI 10.1016/j.jnca.2010.07.006

Tutor: Sanna Suoranta

20. Session management in Cloud Services

Traditional client-server-model of communication had a clear way tomanage sessions for different users. However, original HTTP did nothave state for users in the server side but each page was fetchedseparately. Later, HTTP manages sessions in the server that sends acookie for the browser to store the session identifier in the clientside for future connections. In the cloud services, the web browseris often acting as a very thin client providing only userinterface. How different cloud services enables session management?(Note: hard topic, not much open research available)

References:

1. Subharthi Paul and Jianli Pan and Raj Jain Architectures for the future networks and the next generation Internet: A survey Journal Computer Communications Volume 34 Issue 1, January, 2011 Butterworth-Heinemann Newton, MA, USA doi 10.1016/j.comcom.2010.08.001

Tutor: Sanna Suoranta

Topic by Antti P Miettinen

21. Mobile device energy efficiency issues in HTML5

HTML5 is anticipated to be the technology for future web applicationdevelopment. What are the implications for mobile device battery life?Are there aspects in HTML5 thate enable implementing more energyefficient applications than applications implemented with traditionalAJAX techniques? What are the central challenges in implementing energyefficient HTML5 applications? An example issue in current HTML5specifications is the heated debate about video codecs. H.264 is a codecthat has extensive hardware support and enables therefore energyefficient playback of tag content wheres e.g. Google is heavilylobbying for WebM/VP8. For background information about AJAX energyefficiency issues, see e.g. "A.P. Miettinen, J.K. Nurminen, Analysis ofthe Energy Consumption of JavaScript Based Mobile Web Applications, 2ndInternational Conference on Mobile Lightweight Wireless Systems(MOBILIGHT 2010), May 2010, Barcelona, Spain". This work can be either aliterature study or constructive effort including measurement andanalysis.

References: provided after the topic is assigned.

Tutor: Antti P Miettinen

22. Cloud computing technology for increased battery life

Cloud computing encompasses a plethora of technologies related todistributed computing. What are the technologies that could be used forimproving the battery life of mobile cloud members? How do the differentthin client technologies compare energy wise? Are there significantdifferences in the energy consumption characteristics of differentcommunication protocols? How do different virtualization a storagetechnologies affect mobile client energy consumption? What are the mostpromising technologies for offloading processing from a mobile client tothe cloud? For background information, see e.g. "A.P. Miettinen, J.K.Nurminen, Energy efficiency of mobile clients in cloud computing, 2ndUSENIX Workshop on Hot Topics in Cloud Computing, June 22, 2010, Boston,MA, USA". This work can be either a literature study or constructiveeffort including measurement and analysis.

References: provided after the topic is assigned.

Tutor: Antti P Miettinen

Topic by Zhonghong Ou

23.The Impact of Virtualization on Performance of Cloud Data Centers

As a new paradigm, cloud computing opens new windows of provisioning services and applications by an “on demand” and “pay-as-you-go” manner. However, virtualization of the underlying resources also brings a certain number of tradeoffs. For instance, CPU cores and network interfaces are shared among a bunch of instances, which usually do not belong to the same user. This sharing of resources brings processing latency, incurs unnecessary packet loss, and lowers the network throughput, which are not desirable for cloud users. In this task, you are expected to:

1. Measure the processing delay, packet loss, and TCP/UDP throughput etc, of Amazon EC2 instances
2. Find out the limitations (tradeoffs) of virtualization in cloud data centers

References:

1. Guohui Wang and T.S. Eugene Ng. The Impact of Virtualization on Network Performance of Amazon EC2 Data Center. IEEE Infocom 2010.

Tutor: Zhonghong Ou

24.Does Broadcast-based Address Resolution Protocol (ARP) Work in Data Center Networks?

A subnetwork within a Data Center Network or other dense network may be required to support very large numbers of attached hosts. Alternatively, a region of a network may be required to support significant numbers of subnetworks, where each subnetwork supports significant numbers of hosts. The number of hosts may be particularly large when the hosts are instantiated as virtual machines. Use of conventional broadcast-based ARP (rfc 826) in such subnetworks can result in excessive bandwidth consumption on links associated with the subnetworks. The problem of link bandwidth consumption may be particularly severe in the backbone portion of a network, as in the case of a backbone interconnecting Top of Rack bridges in a Data Center Network, when subnetworks extend across the backbone. Similar to the problem of excessive bandwidth consumption, processing of broadcast-based ARP can place excessive load on the processor associated with a host or a gateway. The problem of excessive processor load is exacerbated when multiple hosts share processor resources, as when the hosts are instantiated as virtual machines. In this assignment, you are expected to:

1. Study the current mainstream solutions for this problem, if any
2. Propose some solution to reduce link bandwidth consumption and processor load to acceptable levels in such environments

Note:
The statement of this problem is copy-pasted from IETF ARMD mailing list. We replicate it here just because it is an interesting practical, meanwhile, research problem.

References:

1. IETF ARMD mailing list. https://www.ietf.org/mailman/listinfo/armd
2. Andy Myers, T. S. Eugene Ng, and Hui Zhang. Rethinking the Service Model: Scaling Ethernet to a Million Nodes. http://www.cs.cmu.edu/~hzhang/papers/hotnets04-ethernet.pdf

Tutor: Zhonghong Ou

Topic by Vilen Looga

25. Survey of MANET-based geosocial networks

Summary:
There are several proposed prototypes of geosocial networks running on top of a MANET, which allow users of mobile devices to discover nearby people and form an ad hoc social network with all the customary services: picture and video sharing, chat, status updates etc etc. Additionally, such networks can provide new kinds of services thanks to their environmental and location awareness: disaster relief, lost children search etc.

Research target:
This paper would survey and analyze currently available systems, describe the technological approach behind them and the services that they provide. The student may also propose new use cases.

References:

1. "PrPl: a decentralized social networking infrastructure" Seok-Won Seong, Jiwon Seo, Matthew Nasielski, Debangsu Sengupta, Sudheendra Hangal, Seng Keat Teh, Ruven Chu, Ben Dodson, and Monica S. Lam. 2 MCS 2010
2. "WiFace: a secure geosocial networking system using WiFi-based multi-hop MANET" Lan Zhang, Xuan Ding, Zhiguo Wan, Ming Gu, and Xiang-Yang Li MCS 2010
3. More prototypes: Nokia Instant Community, One Laptop Per Child, WhereStore

Tutor: Vilen Looga

26. Analysis of offloading use-cases

Summary:
There have been proposed several new frameworks for computational offloading for mobile systems. Their goal is to provide a framework for mobile application developers to offload computationally-heavy parts of their code to a clone image of the operating system running in a cloud, thereby saving energy and speeding up execution. Notable use-cases for such frameworks are image recognition, online debugging, backup etc.

Research target:
So far the use-cases for offloading have been rather weak and most of them have easier solutions based on current technologies. The goal of this paper would be to analyze the most significant proposed use-cases and discuss whether offloading frameworks provide a better solution compared to existing technologies.

References:

1. "Cuckoo: a Computation Offloading Framework for Smartphones" Roelof Kemp, Nicholas Palmer, Thilo Kielmann and Henri Bal MobiCASE 2010
2. "MAUI: making smartphones last longer with code offload" Eduardo Cuervo, Aruna Balasubramanian, Dae-ki Cho, Alec Wolman, Stefan Saroiu, Ranveer Chandra, and Paramvir Bahl. MobiSys 2010
3. "Augmented smartphone applications through clone cloud execution" Byung-Gon Chun and Petros Maniatis. HotOS 2009

Tutor: Vilen Looga

Topic by Andrey Lukyanenko

27. Data Center Traffic Measurements

Data Centers are aggregation points of the network traffic and process, which important as for today's Internet as for the future. Data Centers basically a huge amount of servers which are interconnected, thus as a normal servers they are not free of the problems of instantaneous congestions and overflows.

Here, you are to read a set papers on the existing measurements of the DataCenters traffic and try to get as much as possible what are the problems of current Data Centers, and if possible show how they turn around the difficulties.

References:

1. Benson et al, Network Traffic Characteristics of Data Centers in the Wild, IMC'10.
2. Kandula et al, The Nature of Datacenter Traffic: Measurements & Analysis, IMC'09.

Tutor: Andrey Lukyanenko

28. Client's Puzzles: view of the future

Clients puzzles as an idea were suggested a decade ago. They are mechanisms to prevent Denial-of-Sevice (DoS) attacks on the server, when the clients have to spend some amount of own recourses in order to solve some task, solution of which is “cheap” to check. Not much to add.

However, recently a white-hat hacker, showed that using EC2 cloud it is possible to check a lot of hash primitive per second cheaply. The question is how this finding will affect the whole idea of usage of puzzle solution difficulties?

References:

1. Jules et al, Client puzzles: A cryptographic countermeasure against connection depletion attacks.
2. https://stacksmashing.net/2011/01/12/upcoming-black-hat-talk/

Tutor: Andrey Lukyanenko

29. Fast Wireless (re)connection

Today Wi-Fi connection and reconnection uses seconds of time to scan the network and establish connection to the Access Point.

The question here is to study the process of establishment of connection and try to find the bottleneck in such slow behavior. And ideally, try to suggest how to reduce this slow behavior as much as possible.

References:

1. IEEE 802.11 Standard (2007).
2. Wireless drivers.

Tutor: Andrey Lukyanenko