PROCEEDINGS OF WORLD ACADEMY OF SCIENCE, ENGINEERING AND TECHNOLOGY VOLUME 17 DECEMBER 2006 ISSN 1307-6884

Improving Performance of World Wide Web by

Adaptive Web Traffic Reduction
Achuthsankar S. Nair, and J. S. Jayasudha

technique, intelligent agents are used at the server and client

Abstract—The ever increasing use of World Wide Web in the side to assist them to take intelligent decision on reducing the
existing network, results in poor performance. Several techniques size of the document for each web request.
have been developed for reducing web traffic by compressing the size The remainder of this paper is organized as follows. The
of the file, saving the web pages at the client side, changing the burst related research works for improving network performance by
nature of traffic into constant rate etc. No single method was reducing web traffic are discussed in section 2. Section 3
adequate enough to access the document instantly through the
Internet. In this paper, adaptive hybrid algorithms are developed for
outlines the existing web traffic reduction techniques. Section
reducing web traffic. Intelligent agents are used for monitoring the 4 describes the uses of web traffic reduction. Section 5
web traffic. Depending upon the bandwidth usage, user’s preferences, describes the new adaptive web traffic reduction technique
server and browser capabilities, intelligent agents use the best and algorithms. In section 6, the new adaptive technique is
techniques to achieve maximum traffic reduction. Web caching, incorporated into a browser and its performance is compared
compression, filtering, optimization of HTML tags, and traffic with some existing browsers. Finally, section 7 concludes the
dispersion are incorporated into this adaptive selection. Using this paper.
new hybrid technique, latency is reduced to 20 – 60 % and cache hit
ratio is increased 40 – 82 %. II. RELATED WORKS
A web-based management system for network monitoring is
Keywords—Bandwidth, Congestion, Intelligent Agents, Pre- proposed in [1]. The techniques for improving the network
fetching, Web Caching.
performance by traffic reduction are discussed in [2].
Wisconsin Adaptive Web Assistant (WAWA) is developed in
I. INTRODUCTION
[3] that assist a user in locating specific, current and relevant
T HE exponential rate of growth of World Wide Web
(WWW) has led to an increase in Internet traffic and
degradation in user-perceived latency. With the continuing
information on the World Wide Web. The server side
overhead can be reduced if the number of requested files is
reduced. Multiple files such as HTML text and image files
growth of Internet, web users are suffering from the network
will be packed in an object package file for efficient
congestion and server overloading. Content simplification and
transmission. At the receiving side, a web browser should
data compression techniques are used to reduce the amount of
unpack the files [4]. Knowledge of HTML is not necessary to
bytes sent over the network. Since these techniques cannot be
create a web page. The HTML markup generated by other
used for much reduction in bandwidth consumption, alternate
applications is not of the same standard as hand-coded
schemes such as web caching, web cache sharing, web server
markup. The consequences of such bloated HTML are
pushing, browser initiated server pushing are developed. But
unnecessarily increase storage costs, transmission costs,
these techniques have not yet attained the satisfactory results.
download times and browser rendering times.
Web caching and web cache sharing are unsatisfactory
Optimization of HTML generated by WYSIWYG programs
techniques due to dynamic documents, CPU overhead,
is discussed in [5]. For optimizing web content delivery, web
memory spending for caching, administrative overhead etc.
server accelerator is used in [6]. The web server accelerator
In web server pushing technique, it is difficult for a web
resides in front of a web server and improves server
content provider to know the proper place to push documents.
performance by efficiently delivering cached responses while
In the browser initiated server pushing, the server does not
leaving the role of content generator to the web server. Traffic
know what is in the client cache. Without considering the
dispersion technique is discussed in [7]. In traffic dispersion,
client’s cache, the web server pushes more images than
burst traffic is divided into many sub-bursts that are
needed. The user interacts only with the web browsers to fetch
transmitted in parallel through multiple paths which do not
the web pages. So the browser can store the history of each
share any physical links and are sequenced at the destination.
user behavior and it can take intelligent decision depending
Intelligent browsing for multimedia applications has been
upon the user behavior and traffic condition with the help of
developed in [8]. The learning agents attempts to learn what
intelligent agents. In the proposed adaptive traffic reduction
item the user is searching for by watching the user’s normal
Manuscript received September 29, 2006.
browsing actions. The learning agent builds a profile of the
Achuthsankar S. Nair is working as Hon. Director, Centre for user’s search pattern. The inferred goal provides a reliable
Bioinformatics, University of Kerala, Trivandrum, India. way of estimating the relevance of a multimedia item to the
J. S. Jayasudha is working as Asst. Professor, Department of Computer agent’s actual search goal [8]. A heuristic bandwidth
Science, SCT College of Engineering, Trivandrum, India (phone:
919495376533, 91944311000, e-mail: [email protected]).
allocation method is proposed in [9] for managing the

PWASET VOLUME 17 DECEMBER 2006 ISSN 1307-6884 123 © 2006 WASET.ORG

 PROCEEDINGS OF WORLD ACADEMY OF SCIENCE, ENGINEERING AND TECHNOLOGY VOLUME 17 DECEMBER 2006 ISSN 1307-6884

bandwidth dynamically by considering the correlation can operate on a client or server or on an intermediate system.
characteristics. In proxy caching, proxy server stores cacheable responses to
Cache stores cacheable requests and responses for handling the URL requests, subsequent requests for the same URLs
new requests. If a new request that is same as a stored request yields cache hit. The URL request from the client is forwarded
arrives, then cache can supply the stored response rather than through proxy servers. The proxy servers accept the URL
accessing the resource from the web server. The web caching request from the client and it checks in its own local cache for
and web cache sharing schemes are useful for latency local hit, if requested object is not available there, it forwards
reduction, bandwidth conservation and disconnected that request to the web server. In web cache sharing, for each
operation. If a group of clients are topologically close and URL request, the proxy first checks in its local cache, if there
under common administrative control, then the administrator is no local hit, then that proxy checks in other proxies cache
could install one or more proxy caches in front of the clients for remote hit. If there is no remote hit, then it forwards that
for forming a cluster to lower the client-perceived latency. The request to the web server. Upon receipt of that document, the
methodologies for client cluster identification are discussed in requested proxy server stores it in cache and returns the
[10]. The various web caching techniques, caching protocols document to the client. Due to dynamic documents, CPU
and caching architectures are discussed in [11] - [17]. To overhead, memory spending for caching and administrative
extend the effectiveness of HTTP caches, techniques such as overhead, web caching techniques are considered as
cooperative caching, pre-fetching, partial transfers, delta unsatisfactory techniques. But when the network bottleneck
encoding, cache based compaction and HTML macros are such as congestion is compared, these overhead are negligible.
developed [14, 15]. A detective browser is proposed in [18], it
can immediately determine whether the requested content is D. Web Server Pushing
dynamic or secured. If so, the browser will bypass the proxy Web server pushes some or all of the documents to some
and forward the request directly to the web server. place near the client site. When the client accesses a
document, it will go to a site that is nearer to the client [21].
III. WEB TRAFFIC REDUCTION TECHNIQUES This approach reduces the latency in accessing the documents.
Web traffic reduction techniques are necessary for accessing In this technique, it is difficult for a web content provider to
the web sites efficiently with existing network facility. It is know the proper place to push documents.
costlier to use infinite bandwidth in all organizations. Many
studies show that the web caching has the maximum limit of E. Browser Initiated Server Pushing (BIP)
cache hit ratio of 50%. But pre-fetching can improve the hit In BIP, upon receiving a HTTP request, the server actively
ratio to 60% or even more than 80% [19]. But pre-fetching pushes embedded contents if the permission is given by the
techniques increases the web traffic for pre-fetching the client [21]. By means of the pushing mechanism, the HTML
anticipated sites. Various techniques available for web traffic web page will be down loaded in one RTT if the embedded
reduction are given below. contents and the HTML web page are on the same server. This
approach reduces the download latency for web pages and to
A. Content Simplification improve web server resource utilization. In the browser
Web designers can use common sense to reduce page initiated server pushing, the server does not know what is in
complexity or special tools can be used to optimize image the client cache. Without considering the client cache
coding. But some data such as medical images, broadcast contents, the server pushes much more images than actually
quality videos and executable software cannot be simplified needed.
without loss of meaning. Optimization of HTML tags can be
done for reducing the size of the web page to be transmitted F. Load Balancing Techniques
[5]. But the content simplification and optimization techniques In this approach, if any server is over loaded, its jobs are
cannot be used for much reduction in web traffic. shared by other under loaded servers. Load balancing systems
monitor the health of the available servers and make decisions
B. Compression to route the traffic for optimizing the performance and
Redundant bits within a single transfer can be reduced using availability. This ensures that the users will be connected to
compression techniques. Existing general purpose the most available server, providing excellent and predictable
compression algorithms provide significant size reductions. quality of service to the end user [22]. Many traffic
Several compression techniques such as Huffman coding, LZ, management and load balancing techniques are discussed in
LZW, JPEG, MPEG, H.261, H.263 are available for text, [23]. An adaptive load balancing scheme for web servers is
image, audio and video compression [20]. Compression ratio discussed in [22]. An adaptive multi agent coordination
depends on the compression algorithm and size of the file to algorithm is proposed in [24] for performing distributed
be compressed. These techniques can reduce only temporal or dynamic load balancing. A load cluster management system
spatial redundancy. But it can’t reduce the frequent based on SNMP and web technology is discussed in [25]. A
transmission of the same file through the network. dynamic load balancing algorithm is proposed in [26] for
improving throughput. Even if several load balancing
C. Web Caching algorithms are available, it is difficult for distributing the work
A cache is a facility that stores cacheable requests and among the existing servers and balancing the work among
responses. Subsequent requests can be satisfied from the cache them.
instead of accessing the objects from web server. The cache

PWASET VOLUME 17 DECEMBER 2006 ISSN 1307-6884 124 © 2006 WASET.ORG

 PROCEEDINGS OF WORLD ACADEMY OF SCIENCE, ENGINEERING AND TECHNOLOGY VOLUME 17 DECEMBER 2006 ISSN 1307-6884

G. Intelligent Agents technique and combines the existing techniques such as web
Intelligent agent monitors user actions and dynamically caching, web cache sharing, pre-fetching, traffic dispersion,
takes decision to access web sites by analyzing the traffic on compression etc. Traffic reduction algorithms are developed
the internet. The intelligent agent monitors the web traffic and for combining these techniques. Web traffic reduction is
it reports the status of bandwidth usage to the browser [27]. achieved in adaptive manner by monitoring the user’s
The browser parses the HTML page and the agent makes preferences and bandwidth usage. In this technique, intelligent
prediction about future references. If the traffic reported by the agents monitor the bandwidth usage and select the best
intelligent agents are less than the threshold, then the browser techniques at the client side and server side to achieve
can pre-fetch the predicted references before it is actually maximum traffic reduction. Since it is a hybrid technique,
referenced, potentially resulting in much lower latencies. efficient bandwidth utilization and more web traffic reduction
Using intelligent agents delay in accessing the network, are achieved. Schematic block diagram of the functions of
bandwidth consumption and network bottlenecks can be intelligent agents at web server and browser side is shown in
reduced. Fig. 1.

H. Bandwidth Management A. Web Traffic Reduction at Client Side

Network bottleneck such as congestion can be avoided by The intelligent agents help the browser to make it
managing the bandwidth properly. If infinite bandwidth is intelligent. The user interacts only with the web browsers to
available, there is no need for bandwidth management. It is fetch the web pages. So the browser can store the history of
very costlier for implementing infinite bandwidth in each each user behavior and it can take intelligent decision
organization. If congestion occurs there will be packet loss depending upon the user behavior and traffic condition.
and these packets are to be retransmitted for achieving a Agents at the client side monitor the user’s preferences and
reliable data transfer. Bandwidth management is necessary for bandwidth usage, then takes decision dynamically for
reducing congestion and web traffic. Intra-domain bandwidth reducing latency and web traffic. The client side algorithm for
management in differentiated services network is discussed in web traffic reduction is given below.
[28]. Dynamic bandwidth management scheme for a wireless
network consisting of heterogeneous computers and devices is Traffic Reduction Algorithm at Browser side
proposed in [29]. A two tier resource management model for
the internet is proposed in [30]. An algorithm for bounding the 1. For each URL request
bandwidth of a web crawler is proposed in [31]. a) If the request is secured or dynamic, forward the
request to the web server
I. Intelligent Router b) Else if the request is static, checks in cache
Intelligent router can route the traffic efficiently by (i) If that document is available in browser cache,
reducing the network bottlenecks such as congestion. fetch it from browser cache
Intelligent router is proposed in [32] to implement distributed (ii) Else check in local proxy cache, if local hit,
applications in a heterogeneous environment. Intelligent fetch it from the local proxy cache
routers can dynamically route the web traffic and perform (iii) Else check in remote proxy cache, if remote
reliable service even if some routes are failed. It can take hit, fetch it from that cooperating proxy
dynamic decisions for selecting the route depending on the (iv) Else forward that request to the web server,
current work load on the network. Layer 3 switches are upon receipt of that document, store it in local
available for directing the request for dynamic contents to the proxy cache and browser cache.
web servers and other requests to the cache servers. 2. Intelligent agents monitor the web traffic
a) If bandwidth usage is less than a threshold value say
IV. USES OF WEB TRAFFIC REDUCTION x1, allow all types of traffic
Reduction of latency, reduction of bandwidth usage and b) If bandwidth usage is between threshold values say
alleviating network bottlenecks such as congestion are the x1 and x2, do not allow pre-fetching
main benefits of web traffic reduction. Intelligent agents help c) If bandwidth usage is greater than a threshold values
to access the desired web sites and make the surfing easier. say x2, do not allow pre-fetching and cache
With the help of intelligent agents the web server can refreshing. Display only links of the images on all
negotiate the capabilities of the browser and it can take documents. Fetch images only if the user clicks a
intelligent decision by considering the browsers capabilities particular link for displaying the image.
and traffic condition. By reducing the traffic in the existing 3. When uploading large files, intelligent agents prompt
network, the documents can be accessed in real time. The the user to choose compression techniques.
extra cost that is to be spent for replacing the existing network 4. When searching, the intelligent agent’s identify the
can also be avoided. The existing network is enough to satisfy user’s preferences from log information and display
the user’s request by managing the web traffic efficiently. the related sites first.

V. ADAPTIVE WEB TRAFFIC REDUCTION TECHNIQUE B. Web Traffic Reduction at Server Side
In this new adaptive technique, intelligent agents are The intelligent agents at the server side negotiate with the
maintained at the client and server side for monitoring the web web browser about its capabilities. Depending upon the
traffic. Adaptive web traffic reduction technique is a hybrid capabilities of the browser, the intelligent agents take dynamic

PWASET VOLUME 17 DECEMBER 2006 ISSN 1307-6884 125 © 2006 WASET.ORG

 PROCEEDINGS OF WORLD ACADEMY OF SCIENCE, ENGINEERING AND TECHNOLOGY VOLUME 17 DECEMBER 2006 ISSN 1307-6884

decision about sending the documents in compressed form. (b) If the bandwidth usage is between threshold values
The proposed algorithm for web traffic reduction at server side say x1 and x2, send medium quality image, audio
is given below. and video files.
(c) If bandwidth usage is greater than a threshold
value say x3, send only low quality image, audio
and video files
2. Intelligent agents identify the large files and send it in
compressed form.
3. Intelligent agents separate the static pages from
dynamic pages and make the static pages as public to
cache it in browser side.
4. Intelligent agents identify congestion and reduce the
traffic in that route.
5. Intelligent agents convert the burst nature of the
traffic into sub bursts.
6. Intelligent agents optimize the HTML markup
generated by WYSIWYG programs. Only optimized
HTML is delivered to the browser

VI. SIMULATION AND DISCUSSION OF RESULTS

We have simulated the adaptive traffic reduction technique
by implementing a new browser called dynamic pre-fetching
browser (DPB) with pre-fetching, web caching, traffic
monitoring capabilities. The intelligent agents help the
browser to have these capabilities to reduce latency and to
increase response time. We have analyzed this new technique
by simulating the environment in a medium sized network
containing 90 nodes and trace driven data are collected. We
have compared the percentage bandwidth usage of our
dynamic pre-fetching browser with Internet Explorer (IE) and
Netscape Navigator.
Bandwidth usage (%)

100
80
60
40
20
0
5 10 15 20 25 30
Time (minutes)

DPB IE Netscape

Fig. 2 Percentage use of Existing Bandwidth by Dynamic Pre-

fetching browser (DPB), Internet Explorer (IE) and Netscape
Navigator
Fig. 1 Schematic block diagram of the functions of Intelligent
Agents at Web server and Browser side We have observed that the bandwidth usage is almost
constant using adaptive traffic reduction algorithms by DPB
browser as shown in Fig. 2. Without considering the current
Traffic Reduction Algorithm at Web Server side network traffic, the other browsers attempts to fetch more
objects than the network can afford. But our new technique
1. Intelligent agents monitor the bandwidth usage makes the browser to monitor the current network traffic and
(a) If the bandwidth usage is less than a threshold say user’s preferences and helps the browser to utilize the
x1, send good quality image, audio and video files. maximum bandwidth available effectively. We have observed
that the average bandwidth usage using our new browser is
about 70 to 90 % of the maximum bandwidth available of the

PWASET VOLUME 17 DECEMBER 2006 ISSN 1307-6884 126 © 2006 WASET.ORG

 PROCEEDINGS OF WORLD ACADEMY OF SCIENCE, ENGINEERING AND TECHNOLOGY VOLUME 17 DECEMBER 2006 ISSN 1307-6884

current network. Intelligent agents help the browser to monitor [3] Jude Shavlik, Susan Calcari, Tina Eliassi-Rad and jack Solock, “An
Instructable Adaptive Interface for Discovering and Monitoring
the bandwidth usage and adjust the pre-fetching of subsequent Information on the World-Wide Web”, Proceedings of International
links to make constant traffic and to avoid the network Conference on Intelligent user interfaces, ACM, 1999, pp.157-160.
bottleneck such as congestion. We have also observed that, by [4] Hiroshi Fujinoki, Kiran K Gollamudi, “Object Packaging – Web
incorporating the traffic reduction algorithms in the browser, Response Time Reduction for slow and busy web servers”, Proc. of the
27th Annual Conference on Local Computer Networks, IEEE Computer
latency is reduced at a range of 20 – 60 % and cache hit ratio
society, 2002.
increased 40 – 82 % as shown in Fig. 3 and 4. [5] Jacqueline Spiesser, Less Kitchen, “Optimization of HTML
Automatically generated by WYSIWYG Programs”, Proceedings of the
ACM SIGPLAN 2004 Haskell Workshop, 2004, pp.80-91.
100 [6] Dongjun Shin, Kern Koh, “Optimizing Web Content Delivery using
Web Server Accelerator”, 25th Australasian Computer Science
Cache Hit Ratio

80
Conference, 2002, pp.233-239.
60 [7] Fumio Ishizaki, “Study on reduction of total bandwidth requirement by
traffic dispersion”, In Proceedings of International conference on ATM,
40 IEEE, 2001, pp.285-289.
[8] Chris Drummond, Dan Ionescu, Robert Holte, “Intelligent Browsing for
20
Multimedia Applications”, Proceedings of Multimedia, IEEE, 1996, pp.
0 386-399.
1 3 5 7 9 11 13 15 17 19 21 23 25 [9] Yen-Wen Chen, “Experimental Study of Internet Traffic Modeling and
Bandwidth Allocation”, IEEE Pacific Rim Conference on
Time (Minutes)
Communications, Computers and Signal Processing, 2001, pp.587-590.
[10] Balachander Krishnamurthy, Jia Wang, “On Network – Aware
Adaptive Web Traffic Reduction Technique Web Caching Technique Clustering of Web Clients”, Proceedings of ACM SIGCOMM, 2000, pp.
Fig. 3 Analysis of Cache Hit Ratio by Adaptive Reduction Technique 97-108.
[11] Hossam Hassanein, Zhengang Liang and Patrick Martin, “Performance
and Web Caching Technique Comparison of Alternative Web Caching Techniques”, Procceedings of
the seventh International Symposium on Computers and
Communications, IEEE, 2002.
700 [12] C. Mala and J.S. Jayasudha, “Web Cache Sharing Techniques”,
Proceedings of 6th International Conference of High Performance
Time (Seconds)

600
500 Computing, Asia, 2002.
400 [13] M. Liu, F. Wang, D. Zeng, L.Yang, “An Overview of world wide Web
300 Caching”, International conference on Systems Man and Cybernetics,
200 IEEE, 2001, pp.3045-3050.
100 [14] Greg Barish and Katia, “World Wide Web Caching: Trends and
0 Techniques”, IEEE Communication, May 2000, pp. 178-185.
[15] Jeffrey C. Mogul, “Squeezing More Bits Out of HTTP Caches”, IEEE
1 3 5 7 9 11 13 15 17 19 21 23 25 Network, May/June 2000, pp. 6-12.
No. of Sites [16] Hykyung Bahn, “A Shared Cache Solution for the Home Internet
Gateway”, IEEE Transactions on Consumer Electronics, Vol.50, No.1,
Adaptive Web Traffic Reduction Technique Feb 2004, pp. 168-172.
Without Adaptive Reduction Technique [17] Li Xiao, Xiaodong Zhang, Artur Andrzejak, Songqing Chen, “Building a
Large and Efficient Hybrid Peer to Peer Internet Caching System”, IEEE
Fig. 4 Analysis of Latency by Adaptive Reduction Technique and Transactions on Knowledge and Data Engineering, Vol 16, No.6, June
Web Caching Technique 2004, pp.754-769.
[18] Songqing Chen and Xiaodong Zhang, “Detective Browsers : A Software
Technique to Improve Web Access Performance and Security”,
Proceedings of 7th International workshop on Web content caching and
VII. CONCLUSION Distribution (WCW’ 02), 2002.
The intelligent agents help the browser and web server to [19] Shi L, Gu Z, Wei L and Shi Y. “Popularity-based selective markov
model”. In Proceedings of the IEEE/WIC/ACM International
monitor the bandwidth usage and user’s preferences. In this Conference on Web Intelligence, 2004, pp. 504-507.
paper an adaptive technique is proposed for reducing web [20] Fred halsall, Mutimedia Communications, Pearson Education, 2002.
traffic and to access the web sites efficiently. The proposed [21] Wenting Tang, Matt W. Mutka, “Intelligent Browser Initiated Server
algorithms at client side and server side are efficient to reduce Pushing”, IEEE International Conference on Performance,
Computing, and Communications Conference, 2000, pp.17-23.
the web traffic in adaptive manner. The simulation results [22] James Aweya, M. Ouellette, D.Y Montuno, B. Doray, K. Felske, “An
show that this adaptive technique maintains almost constant Adaptive Load Balancing Scheme for web servers”, International
traffic and provides effective bandwidth usage. Since it is a Journal of Network Management, 2002, pp.3-39.
hybrid technique, latency is reduced to 20 – 60 % and cache [23] http://www.zenus.com/solutions/zlb/indexfull.html
[24] David Vengerov, Hamid R Berneji, Alex Vengerov, “An adaptive
hit ratio is increased 40 – 82 %. New technologies are to be coordination among fuzzy reinforcement learning agents performing
developed to reduce web traffic without increasing the cost distributed dynamic load balancing”, Proceedings of 11th IEEE
spent for networking. International Conference on Fuzzy Systems, 2002, pp.179-184.
[25] Myung-Sup kim, Mi-Jeong Choi, James W. Hong, “Highly available
and efficient load cluster management system using SNMP and web”,
REFERENCES Proceedings of Network Operations and Management Symposium,
[1] Jae-Won Choi and Kwang-Hui Lee, “A Web-based Management System IEEE, 2002, pp.619-632.
for Network Monitoring”, IEEE workshop on IP operations and [26] Hemant B. More, Jie Wu, “Throughput Improvement through Dynamic
management (IPOM 2002), 2002, pp. 98-102. Load Balance”, IEEE southeast con’94, 1994, pp.339-342.
[2] Zhen Sheng Guo, Yan Zhuang, “Improving Network Performance by [27] Jayasudha J.S, Achuthsankar S. Nair, “An Intelligent Browser for Web
Traffic Reduction”, International Conference on Information, Traffic and Latency Reduction”, Proceedings of International
Communication and Signal Processing, 1997, pp. 1226-1230.

PWASET VOLUME 17 DECEMBER 2006 ISSN 1307-6884 127 © 2006 WASET.ORG

 PROCEEDINGS OF WORLD ACADEMY OF SCIENCE, ENGINEERING AND TECHNOLOGY VOLUME 17 DECEMBER 2006 ISSN 1307-6884

Conference of Trends in Industrial Measurements and Automation

(TIMA), 2004.
[28] S. Jha, M. Hassan, P.Nanda, N. Ahmed, “Intra-domain Bandwidth
Management in Differentiated Services Network”, IEEE Conference on
Local Computer Networks, 2000, pp.326-327.
[29] Samarth H. shah, K. Chen, K. Nahrstedt, “Dynamic Bandwidth
Management for single-hop Ad-hoc Wireless Networks”, Proceedings of
first IEEE international Conference on Pervasive Computing and
Communications, 2003.
[30] A. Terzis, L. Wang, J. Ogava, L. Zhang, “A two-tier Resource
Management Model For The Internet”, Proceedings of Global
Telecommunications Conference, IEEE, 1999, pp. 1779-1791.
[31] M. Diligenti, M. Maggni, F.M. Pucci, F. Scarselli, “Design of a Crawler
with bounded bandwidth”, Proceedings of WWW-2004, ACM, 2004, pp.
292-293.
[32] C. Daniel Wolfson, Ellen M. Voorhees, M.M. Flatley, “Intelligent
Routers”, Proceedings of 9th International Conference on Distributed
Computing Systems, IEEE Computer Society, 1989, pp. 371-376.

PWASET VOLUME 17 DECEMBER 2006 ISSN 1307-6884 128 © 2006 WASET.ORG