DSpace at VNU: Mobile Peer-to-Peer Approach for Social Computing Services in Distributed Environment

This approach features peer-to-peer networks with the support of mobile devices that allows users to par-ticipate in social computing services including data shar-ing and searchshar-ing

Mobile Peer-to-Peer Approach for Social Computing

Services in Distributed Environment

Ha Manh Tran School of Computer Science and Engineering International University Vietnam National University tmha@hcmiu.edu.vn

Khoa Van Huynh Department of Network and Service Management VNPT Dong Thap VNPT Group khoahv.dtp@vnpt.vn Khoi Duy Vo

School of Computer Science and Engineering International University Vietnam National University khoivd@hcmiu.edu.vn

Son Thanh Le School of Computer Science and Engineering International University Vietnam National University ltson@hcmiu.edu.vn ABSTRACT

This paper presents a mobile peer-to-peer approach that can

be applied to build social computing services in distributed

environment This approach features peer-to-peer networks

with the support of mobile devices that allows users to

par-ticipate in social computing services including data

shar-ing and searchshar-ing services Due to the limitations of

sys-tem resource and network connectivity, mobile peers cannot

easily undertake complicated operations, such as processing

complex queries, indexing and transmitting large amount

of data This approach employs a super peer peer-to-peer

network to deal with the problem of peer heterogeneity It

uses workstations as peers to assist mobile peers with

insuf-ficient storage, bandwidth and processing capability in

deal-ing with complicated operations, while mobile peers possess

ordinary operations such as publishing and searching data

We have extended the Gnutella protocol to provide

oper-ations on peers and mobile peers The evaluation of the

prototyping system has performed on a number of

labora-tory workstations and Android emulators to investigate the

feasibility and scalability of the system

Categories and Subject Descriptors

C.2 [Computer Communication Networks]: Distributed

Systems—peer-to-peer networks, decentralized online social

networks, mobile applications; H.3.3 [Information Search

and Retrieval]: [software bug report retrieval]

Keywords

Peer-to-Peer Networks, Decentralized Online Social Networks,

Permission to make digital or hard copies of all or part of this work for

personal or classroom use is granted without fee provided that copies are

not made or distributed for profit or commercial advantage and that copies

bear this notice and the full citation on the first page To copy otherwise, to

republish, to post on servers or to redistribute to lists, requires prior specific

permission and/or a fee.

SoICT’13 December 05-06, 2013, Danang, Vietnam

Copyright is held by the owner/author(s) Publication rights licensed to

ACM ACM 978-1-4503-2454-0/13/12 $15.00.

http://dx.doi.org/10.1145/2542050.2542064

Social Computing Services, Mobile Computing Services, Soft-ware Bug Report Retrieval

Content delivery is the tendency of today data commu-nication Social networks, peer-to-peer (P2P) networks and virtual communities, such as Youtube video sharing, Face-book social network, BitTorrent P2P network contribute a significant amount of digital content to the Internet These networks also attract a huge number of users participat-ing in online services, such as online games, multimedia file searching and sharing, video and audio streaming With the growth of mobile and networking technology, users tend to use mobile devices for these online services on the Internet Mobile computing has been recently applied to social net-works, P2P networks and virtual communities, where mobile devices not only use services but also undertake operations related to network formation and service provision

Applying mobile devices to the complicated operations can be demanding, but challenging Mobile devices contain

a number of limitations of network connectivity, system re-source, peripherals and power consumption The networks using mobile devices possibly face several problems of inter-mittent connectivity, processing incapability and data un-availability, thus reducing the performance and reliability of the networks We propose a mobile P2P approach that em-ploys an appropriate P2P network with the support of mo-bile devices to foster user online activities This approach first applies the super peer P2P network to support mo-bile devices as momo-bile peers for network formation This network combines peers and mobile peers to assist peers with limited storage, bandwidth and processing capability

in dealing with complicated operations, such as processing complex queries, indexing large databases, or transmitting a large amount of data The approach then exploits the data sharing and searching capability of P2P networks to provide social computing services on distributed environment The contribution is thus threefold:

1 Studying the existing P2P protocols and applications supporting for mobile devices, then proposing a

mo-bile P2P approach with network formation and search

service

2 Implementing several components of the prototyping

system including query handler and data publisher with

the extension of the Gnutella protocol

3 Evaluating the feasibility and scalability of the

proto-typing system using software bug report datasets, such

as RedHat, Mozilla, Ubuntu, Asterisk, etc

The rest of the paper is structured as follows: the next

sec-tion includes some background of P2P networks and mobile

P2P networks, describes the existing P2P frameworks for

building social networks and presents the requirement

anal-ysis of a feasible mobile P2P network Section 3 provides a

study of the existing mobile P2P protocols and the recent

de-velopment of mobile P2P applications that help choosing an

appropriate P2P protocol and development platform

Sec-tion 4 proposes a mobile P2P approach with the detailed

de-sign of system architecture, component communication and

protocol support Several experiments in Section 5 report

the feasibility and scalability of the prototyping system

be-fore the paper is concluded in Section 6

A P2P network contains a large number of workstations

that share computing resources including storage, bandwidth

and processor power Each workstation or peer in the

net-work acts as a client and server to consume and provide

ser-vices respectively Peers can dynamically join and leave the

network without causing the instability of other peers The

network also possesses advantages in reducing collaboration

cost through ad-hoc communication process and providing

high fault-tolerance and scalability There are three types

of P2P networks: structured, unstructured and hybrid

net-works

The structured P2P network is tightly controlled in

topol-ogy and a peer is fixed in a logical location when connecting

to other peers This kind of networks uses Distributed Hash

Table (DHT) to generate uniquely consistent identifiers for

peers and resources such that the peers hold the resource

indexes if their identifiers are in the same identifier space

Lookup queries are forwarded to the peers which are closer

to the resources in the identifier space The prevailing

struc-tured P2P systems are CAN [16], Chord [18], Kademlia [15],

The unstructured P2P network is loosely controlled in

topology and a peer connects to other peers in a random

fashion Each peer maintains a list of resources in the local

repository Flooding-based search is a common mechanism

used to find resources in this kind of networks Peers send

queries to the neighboring peers for queryhits The key

dis-advantage of these networks is severe scalability problem

as the number of queries and peers increase The

prevail-ing unstructured P2P systems are Gnutella [2], Freenet [6],

BitTorrent [7] The super peer P2P network is a hybrid

network that combines the characteristics of the P2P

net-work with the client-server netnet-work to address the problem

of heterogeneous peers, i.e., peers possess various capability

of storage, bandwidth and processing power The study of

Yang et al [23] has presented guidelines for designing the

super peer network to take advantage of peer capabilities

The super peer network comprises many clusters connected

to each other to form either structured or unstructured P2P

networks, in which each cluster contains a super peer and

a set of clients The clients submit queries to, and also ob-tain queryhits from, their super peer while the super peers forward the queries and receive the queryhits on the super peer network The latest version of the Gnutella protocol has included this super peer concept

The study of Hu et al [12] has proposed an architecture that uses mobile agents on Gnutella networks to support file sharing capability for mobile devices This study has used the JXTA framework [1] to implement the architecture However, our study applies mobile computing on P2P net-works to build data sharing and searching services on mobile P2P networks It emphasizes data search capability of mo-bile P2P networks for social computing services, rather than only file sharing capability A mobile P2P network is a P2P network with both mobile devices and workstations acting

as peers Differences between a mobile device and a worksta-tion cause several differences between a mobile P2P network and a P2P network First, mobile devices face a problem of intermittent network connectivity due to battery exhaus-tion, network latency and low bandwidth, signal coverage and mobility Second, mobile devices possess limited system resources including less memory, weak processors and small storage compared to workstations Third, mobile devices contain limited input and output devices With these dis-advantages, mobile peers cannot easily perform complicated operations on P2P networks, e.g., the shared data on mobile peers becomes unreliable due to intermittent connectivity; mobile peers become a bottleneck due to the incapability of processing queries To provide a feasible mobile P2P net-work, the problems of data availability and peer capability need to be addressed Moreover, mobile devices also need an application development environment with the full support

of libraries, e.g., Android platform

Decentralized online social networks provide various so-cial computing services on distributed environment These services improve the limitations of the centralized servers by using the decentralized servers Several studies have applied P2P technology to building decentralized social networks Safebook [8] adopts a decentralized architecture relying on cooperation among users to deal with the user privacy and provider application It also builds trust relationship be-tween users for using online applications LifeSocial [11, 10] tackles the problem of high administration cost on multi-media online communities that heavily depend on the cen-tralized systems It uses a P2P framework to construct a network architecture and extend functions to include social networking services, such as user profiles, friend lists, groups, photo albums, etc This framework has later been applied for building secure online social networks PeerSoN [5] re-solves the problem of user privacy and Internet connectivity

on social networks It replaces the centralized authority of social networks and provides direct data communication be-tween network nodes by using a P2P framework with encryp-tion It also use distributed storage to foster local services Some of the studies have provided the prototyping systems and API interfaces for implementing services

APPLI-CATIONS

The study of Almudena et al [9] has included the overview

of several existing mobile P2P protocols Several

proto-Table 1: Overview of mobile P2P protocols (as of Jan 2013)

Protocols Platform License Update

BitTorrent Multiple No 2013 Proprietary∗ Multiple Optional 2013

cols along with frameworks have been proposed for years,

such as Proem [13], JMobiPeer [4], Mobile Chedar [14] and

Peer2Me [22] These protocols share common characteristics

of being without license and using the Java/J2ME platform

The Symella protocol [3] is a Gnutella-like protocol proposed

for file sharing applications on the Symbian platform Only

few applications have found on these protocols According

to our study, BitTorrent protocol [7] and some proprietary

protocols have been heavily used for mobile P2P

applica-tions The BitTorrent protocol contains several advantages

used for file sharing applications on both workstations and

mobile devices A centralized server maintains torrent files

that track peers participating in torrent, while downloading

files occur on peers The proprietary protocols apply

exist-ing protocols such as SIP signallexist-ing protocol [17] to VoIP

applications These protocols can easily be implemented on

multiple platforms of mobile devices Table 1 presents the

update of the existing mobile P2P protocols as of January

2013 Items marked with * mean unpublished protocols

The existing mobile P2P protocols contain multiple

re-markable features We focus on exploiting the resource of

mobile devices not only for file sharing applications but

also mobile search applications Table 2 reports the

shar-ing and searchshar-ing capability of the above protocols Except

for the proprietary protocols used for VoIP applications,

other protocols support file sharing features All

proto-cols ignore sharing computing resources, thus making them

hard to support computing applications that heavily use

sys-tem resource on mobile devices Some protocols such as

Proem, JMobiPeer, Mobile Chedar, Gnutella, Symella and

BitTorrent provide basic search capability that allow peers

to search file names by matching keywords The Mobile

Chedar and BitTorrent protocols also use servers to support

sharing and searching operations on peers Among several

protocols, the Mobile Chedar protocol can be most suitable

for mobile search applications except for one limitation of

using Bluetooth technology for communication

The BitTorrent protocol has been widely applied to file

sharing applications Table 3 describes an incomplete list of

mobile P2P implementations based on the BitTorrent

proto-col The Android OS seems to be a good platform for mobile

applications because a large number of open source

imple-mentations run on the Android OS These impleimple-mentations

have also attracted a large number of users

Figure 1 plots the architecture of the proposed mobile P2P

system that focuses on data sharing and searching services

Note that the architecture can also be extended to covering

specific social computing services There are two kinds of

peers in this system: mobile peer and search peer Consid-ering several limitations of mobile peers, search peers are designed to assist mobile peers’ operations while they are offline due to connection problems, resource constraint or battery exhaustion This architecture coincides the archi-tecture of the super peer P2P networks [23], where search peers acts as super peers, and mobile peers are referred to

as incapable peers, i.e., peers have insufficient capability to perform complicated operations The key difference is that search peers support searching facility by storing, locating and indexing data for mobile peers, while mobile peers fos-ter sharing facility by publishing all data or a part of data

on search peers

Figure 1: Mobile P2P system topology With this system, mobile peers connect to at least one search peer when they get online There exist bootstrap-ping search peers available on the network Mobile peers advertise their data and possibly upload a part of their data

to search peers Therefore, searching data only occurs on search peers, while retrieving data occurs on both search peers or mobile peers Mobile peers send queries to search peers, and then search peers in turn forward the queries to other search peers The queryhits with either the result-ing data or the resultresult-ing mobile peers are responded to the querying mobile peers Mobile peers can also connect to the resulting mobile peers for downloading the resulting data Figure 2 plots the communication of peers and components

in the system Search peers and mobile peers possess the same components: peer controller, query handler, data pub-lisher and database However, the functionality of these components can be different on each kind of peers

Peer controller manages communication among peers and components This component exchanges information to both

Table 2: Overview of sharing and searching features of mobile P2P protocols (as of Jan 2013)

Protocols Multimedia Computing Search Server

Resource Resource Capability Support

peers and components depending on several types of

mes-sages, e.g., a query is forwarded to the query handler

com-ponent or a request of publishing data is forwarded to the

data publisher component Query handler is responsible for

query processing Upon receiving a query, this component

obtains and processes relevant data from the database and

returns the queryhit This component requires processing

and storage capability, thus mobile peers can only use this

component for downloading the results of the queryhits and

processing very limited database Data publisher maintains

data items and references from mobile peers This

compo-nent creates indexes from the database for search methods

Mobile peers use this component to keep track on the

pub-lished data

Figure 2: Communication of peers and components

in the mobile P2P system

The Gnutella P2P protocol [2] is most suitable for this

system due to several reasons First, this protocol supports

super peers for solving the problem of incapable peers, e.g.,

super peers are responsible for routing queries Second, this

unstructured and flooding-based protocol provides flexible

query processing, thus facilitating both keyword and

seman-tic search methods Third, this protocol also bolsters data

replication for better data availability Fourth, there are

multiple open source implementations of this protocol The

applicability of this protocol to the system is straightforward

because this protocol has already been extended for the P2P

search systems [21, 20] More specifically, the Gnutella

mes-sages need to be changed to enable more efficient search

methods on search peers and allow mobile peers to upload

data before going offline The Gnutella protocol supports five types of messages: ping and pong used to probe the net-work, query and queryhit used to exchange data, and push used to deal with peers behind the firewall Downloading data is handled separately from this protocol A Gnutella message consists of header and content The attributes of the header are shown as follows:

Original Gnutella message header + -+ -+ -+ -+ -+

| message id | descriptor | ttl | hops | payload length | + -+ -+ -+ -+ -+

The message id field is used to detect whether a message has already arrived at a certain peer before The payload descriptor field indicates the type of a message such as, ping (0×00), pong (0×01), query (0×80), queryhit (0×81) and push (0×40) The ttl field is the number of times that a message can be forwarded in the network while the hops field

is the number of times that a message has been forwarded The payload length field is the in-byte size of the content that immediately follows the header The detailed structure

of the content depending on message types is defined in the protocol specification [2]

The ping and query messages remain unchanged, while the pong and queryhit messages are extended to carrying the publishing data and the resulting data exchanging between

a mobile peer and a search peer, respectively The attributes

of the pong and queryhit messages are shown as follows:

Original Gnutella pong message + -+ -+ -+ -+ -+

| port | ip addr | num files | num kb | optional data | + -+ -+ -+ -+ -+ Original Gnutella queryhit message

+ -+ -+ -+ -+ -+ -+

| num hits | port | ip addr | speed | result set | servent id | + -+ -+ -+ -+ -+ -+

Upon receiving the ping message from a search peer, a mobile peer uses the pong message to advertise the publish-ing data through number of file shared, number of kilobytes shared and optional data fields Upon responding the query message from a mobile peer, a search peer uses the query-hit message to include the resulting data through speed and result set fields The information of port and IP address specifies the downloaded peers The information of mini-mum speed and speed discloses the data transmission rate

of the downloaded peers

Table 3: Overview of mobile P2P implementations based on the BitTorrent protocol (as of Jan 2013)

Implementations License OS Update Num of Users

WinMobile Torrent Shareware Windows mobile 2013 90549

Swarm Torrent Client Shareware Android 2013 39675 RuTracker Downloader Freeware Android 2013 75864 Torrent Downloader Freeware Android 2013 124675

uTorrent Remote Freeware Android 2013 10483

We have used a Java open source implementation of the

Gnutella protocol that can run on workstations We have

extended this implementation to run on Android emulators

with data sharing and searching capability Note that the

Gnutella protocol basically supports simple keyword based

search method, while our system aims at providing more

complicated search methods We have used software bug

datasets [19] to evaluate data sharing and searching

opera-tions on the mobile P2P system Bug reports are crawled

from several bug tracking systems, such as Bugzilla RedHat,

Bugzilla Mozilla, Launchpad Ubuntu, Asterisk Mantis, etc

A bug report usually contains meta-data for administrative

activity and text description for problem solving, which

al-low the system to provide both keyword and semantic search

methods The experiments mainly focus on the feasibility

and scalability of the system In particular, we evaluate

memory usage, network traffic generation, data availability

and response time on the system

0

50

100

150

200

250

20 40 60 80 100

Number of Bug Reports (x1000) Search peer

Figure 3: Memory usage for different bug datasets

on search peers

Search peers are required to possess sufficient resource

ca-pability, while mobile peers play a role of incapable peers

The first experiment measures the memory usage of a peer

for different bug datasets Figure 3 shows that a peer needs

approximately 200 MB RAM memory to store 100.000 bug

reports This memory usage is reasonable for a workstation

with a normal hardware configuration A mobile peer

possi-bly uses approximately 10 MB RAM memory to load a small

number of 5.000 bug reports Note that bug reports are

var-ious in size ranging from few kilobytes to few megabytes

A Gnutella peer in the Gnutella network obtains the

re-sulting data indexes in the queryhit messages and then

0 2 4 6 8 10

20 40 60 80 100

Number of Queries

Gnutella peer Search peer Mobile peer

Figure 4: Average traffic generation comparison for the Gnutella peer, the search peer and the mobile peer

forms downloading the resulting data from the other peers The download process is separate from the query process that usually generates a lot of network traffic on the network

A search peer in the proposed system uses the queryhit mes-sages to carry the resulting data except for situations that mobile peers refuse to publish data to peers Publishing data to peers improves data availability due to the inter-mittent connectivity problem of mobile peers The network traffic generated by peers therefore increases on the network considerably as the number of queries increases The sec-ond experiment compares the average traffic generated by the Gnutella peer, the search peer and the mobile peer us-ing different query sets, as shown in Figure 4 We have set 30 most relevant bug reports for each queryhit message and the content of each bug report significantly contributes the size of the message while the Gnutella queryhit message only specifies short file names A mobile peer only sends queries, receives queryhits and ignores the majority of the query process, thus generating less network traffic

Mobile peers can choose to publish all data, a part of data

or no data on search peers Data availability thus depends

on the existence of mobile peers, given an assumption that search peers are rarely offline The third experiment com-pares three situations using different churn rates Churn rate presents the number of peers moving out the network over a certain period of time We have tested on 20 mobile peers with mobile peer rate set to 10% to 50% (or 1 to 5), e.g., with the mobile peer rate of 10%, there are 2 mobile peers publishing no data, 2 mobile peers publishing 50% of

0

50

100

150

200

250

0 1 2 3 4 5

Mobile Peer Rate

Churn rate 10%

Churn rate 40%

Figure 5: Data availability for different sets of

mo-bile peers

data, and the remaining mobile peers publishing all data

Each mobile peer contains a bug dataset of 5.000 bug

re-ports that provide approximately 10 MB data at maximum

Figure 5 shows that data availability slowly reduces as the

mobile peer rate increases With the churn rate of 40% and

the mobile peer rate of 50%, the amount of data published

on the network reduces approximately 50% Note that data

is unavailable because mobile peers refuse to upload data to

peers before going offline Replication can help resolving the

problem

0

50

100

150

200

2 3 4 5 6 7 8 9

Time Consumption (s)

Search peer Mobile peer

Figure 6: Time consumption for obtaining number

of bug reports on search peers and mobile peers

The system requires fast response time that cannot be

easy to achieve on the P2P networks because of the

de-pendence of peer capability The fourth experiment

mea-sures the time consumption of the query process for

differ-ent datasets, as shown in Figure 6 A mobile peer sends

each query to its search peer that in turn forwards to other

search peers and also replies to the mobile peer if it possesses

data relevant to the query Time consumption is therefore

measured by recording wall time on search peers and mobile

peers when they obtain queryhits Each queryhit contains

30 relevant bug reports at maximum With the same period

of time, mobile peers receive fewer bug reports than search

peers, e.g., mobile peers and search peers receive 40 and 60

bug reports respectively after 5 seconds Response time to

mobile peers is higher than search peers using a laboratory

network

We have proposed the mobile P2P approach that can be

applied to building social computing services on distributed

environment, particularly data sharing and searching ser-vices This approach allows mobile devices to undertake complicated operations including data sharing and search-ing operations on P2P networks Due to the limitations of system resource and network connectivity of mobile devices, the architecture design of the proposed mobile P2P system needs to consider: (i) search peers with storage and pro-cessing capability that can perform searching operations, (ii) mobile peers with data publishing capability that can improve data availability, and (iii) both keyword and se-mantic search methods We have found that several exist-ing mobile P2P protocols do not focus on data sharexist-ing and searching on P2P networks using mobile peers We have implemented the prototyping mobile P2P system with the above requirements The system recruits the Gnutella P2P protocol where peers possess sufficient processing and stor-age capability to accommodate large databases and perform queries on the databases, while mobile peers can only publish and download relevant data The experiments evaluate the feasibility and scalability of the system using several met-rics: memory usage, network traffic generation, data avail-ability and response time The experimental results reveal that while search peers use more memory and generate more network traffic, mobile peers still use reasonable system re-source Moreover, data availability and time response are acceptably affected by the leaving of mobile peers due to intermittent connectivity Future work considers the possi-bility of extending the mobile P2P approach to other social computing services including distributed user profile storage and information sharing group

This research work is funded by Vietnam National Foun-dation for Science and Technology Development (NAFOS-TED) under grant number 102.02-2011.01

[1] JXTA Project http://jxta.kenai.com/ Last access in Mar 2012

[2] Gnutella Protocol Specification version 0.4 http://rfc-gnutella.sourceforge.net/developer/stable/index.html,

2001 Last access in Mar 2013

[3] Symella http://amorg.aut.bme.hu/projects/symella,

2006 Last access in Jan 2013

[4] M Bisignano, G D Modica, and O Tomarchio JMobiPeer: A Middleware for Mobile Peer-to-Peer Computing in MANETs In Proc 1st International Workshop on Mobility in Peer-to-Peer Systems (ICDCSW ’05), pages 785–791, Washington, DC, USA, 2005 IEEE Computer Society

[5] S Bucheggerand, D Schi´’oberg, L H Vu, and

A Datta PeerSoN: P2P Social Networking: Early Experiences and Insights In Proc 2nd ACM EuroSys Workshop on Social Network Systems (SNS ’09), pages 46–52 ACM, 2009

[6] I Clarke, O Sandberg, B Wiley, and T W Hong Freenet: A Distributed Anonymous Information Storage and Retrieval System In Proc International Workshop on Design Issues in Anonymity and Unobservability, pages 46–66, Heidelberg, Germany,

2000 Springer-Verlag

[7] B Cohen Incentives Build Robustness in Bittorrent.

In Proc 1st Workshop on Economics of Peer-to-Peer

Systems, 2003

[8] L A Cutillo, R Molva, and T Strufe A

Privacy-Preserving Online Social Network Leveraging

on Real-life Trust Communications Magazine,

47(12):94–101, 2009

[9] A D´ıaz, P Merino, L Panizo, and A M Recio A

Survey on Mobile Peer-to-Peer Technology In Proc

XV Conference on Concurrency and Distributed

Systems (JCSD ’07), pages 59–68, Nov 2007

[10] K Graffi, C Gross, D Stingl, D Hartung,

A Kovacevic, and R Steinmetz LifeSocial.KOM: A

Secure and P2P-based Solution for Online Social

Networks In Proc IEEE Consumer Communications

and Networking Conference (CCNC ’11) IEEE

Computer Society, 2011

[11] K Graffi, S Podrajanski, P Mukherjee, A Kovacevic,

and R Steinmetz A Distributed Platform for

Multimedia Communities In Proc 10th IEEE

International Symposium on Multimedia (ISM ’08),

pages 208–213 IEEE Computer Society, 2008

[12] T H Hu, B Thai, and A Seneviratne Supporting

Mobile Devices in Gnutella File Sharing Network with

Mobile Agents In Proc 8th IEEE Symposium on

Computers and Communications (ISCC ’03), pages

1035–1040 IEEE Computer Society, 2003

[13] G Kortuem, J Schneider, D Preuitt, T Thompson,

S Fickas, and Z Segall When Peer-to-Peer comes

Face-to-Face: Collaborative Peer-to-Peer Computing

in Mobile Ad hoc Networks In Proc 1st International

Conference on Peer-to-Peer Computing (P2P ’01),

pages 75–, Washington, DC, USA, 2001 IEEE

Computer Society

[14] N Kotilainen, M Weber, M Vapa, and J Vuori

Mobile Chedar–A Peer-to-Peer Middleware for Mobile

Devices In Proc 3rd IEEE International Conference

on Pervasive Computing and Communications

Workshops (PERCOMW ’05), pages 86–90,

Washington, DC, USA, 2005 IEEE Computer Society

[15] P Maymounkov and D Mazi`eres Kademlia: A

Peer-to-Peer Information System Based on the XOR

Metric In Proc 1st International Workshop on

Peer-to-Peer Systems (IPTPS ’01), pages 53–65,

London, UK, 2002 Springer-Verlag

[16] S Ratnasamy, P Francis, M Handley, R Karp, and

S Schenker A Scalable Content Addressable Network

In Proc Conference on Applications, Technologies,

Architectures, and Protocols for Computer

Communications (SIGCOMM ’01), pages 161–172,

New York, NY, USA, 2001 ACM Press

[17] J Rosenberg, H Schulzrinne, G Camarillo,

A Johnston, J Peterson, R Sparks, M Handley, and

E Schooler SIP: Session Initiation Protocol RFC

3261 (Informational), June 2002

[18] I Stoica, R Morris, D Karger, M F Kaashoek, and

H Balakrishnan Chord: A Scalable Peer-to-Peer

Lookup Service for Internet Applications In Proc

Conference on Applications, Technologies,

Architectures, and Protocols for Computer

Communications (SIGCOMM ’01), pages 149–160,

New York, NY, USA, 2001 ACM Press

[19] H M Tran, G Chulkov, and J Sch¨onw¨alder

Crawling Bug Tracker for Semantic Bug Search In Proc 19th IFIP/IEEE International Workshop on Distributed Systems: Operations and Management, pages 55–66 Springer-Verlag, 2008

[20] H M Tran, S T Le, S V U Ha, and T K Huynh Software Bug Ontology Supporting Bug Search on Peer-to-Peer Networks In Proc 6th International KES Conference on Agents and Multi-agent Systems – Technologies and Applications (AMSTA ’13) IOS Press, 2013

[21] H M Tran and J Sch¨onw¨alder Heuristic Search using a Feedback Scheme in Unstructured Peer-to-Peer Networks In Proc 5th International Workshop on Databases, Information Systems and Peer-to-Peer Computing Springer-Verlag, 2007 [22] A Wang and K Saxlund Peer2Me - Rapid Application Framework for Mobile Peer-to-Peer Applications In Proc International Symposium on Collaborative Technologies and Systems (CTS ’07) IEEE Press, 2007

[23] B Yang and H Garcia-Molina Designing a Super-Peer Network In Proc 19th International Conference on Data Engineering (ICDE’03), page 49, Los Alamitos, CA, USA, 2003 IEEE Computer Society