Web Access In Pervasive Environments- A Review

Web Access in Pervasive Environments: A Review Roushdat Elaheebocus School Of Electronics and Computer Science University of Southampton, UK [email protected] Abstract. In recent years there have been considerable research work in the area of Web Access through mobile devices. In this review paper, five important papers related to this area have been analysed. It has been uncovered that most solutions are proxybased although hybrid versions exist. We also found that security and scalability aspects of the solutions have often been ignored and therefore most of them are not practical in real-life scenarios.

Keywords: web access, pervasive, ubiquitous

Table of Contents 1 Introduction & A historical overview .........................................................................................2 2 Papers' Summary ........................................................................................................................2 2.1 Paper 1..................................................................................................................................2 2.2 Paper 2..................................................................................................................................2 2.3 Paper 3..................................................................................................................................2 2.4 Paper 4..................................................................................................................................2 2.5 Paper 5..................................................................................................................................3 3 Critical and comparative evaluation............................................................................................3 4 Conclusion...................................................................................................................................4 5 References....................................................................................................................................4 6 Bibliography................................................................................................................................5 7 Appendices ..................................................................................................................................5 7.1 Appendix A: Selection Process............................................................................................5 7.2 Appendix B. How this report influenced my ideas..............................................................6 7.3 Appendix C. Lessons learnt.................................................................................................6

Web Access in Pervasive Environments: A Review

1/7

1 Introduction & A historical overview With Mark Weiser's 1991's vision about technologies that are embedded into our everyday life whereby services can be accessed anywhere and at any time [1], the foundation was set for ubiquitous computing environments. Around the same period, T. Berners-Lee proposed a Hypertext project described as the World Wide Web [2] or the Web for short. Roughly three years later, researchers started to come up with ways of accessing the Web using electronic devices over wireless networks and thus successfully bridging the above-mentioned two research areas. The incompatibility of HTML with PDA displays were discussed along with the pros and cons of using such devices to access information online [3].An application programming interface was even developed to help in the design and implementation of applications for multi-device wireless networks [4]. Mobisaic [5] explored the idea of using contextual information about mobile clients and how information can be adapted through the use of dynamic URL. In 1995, the Wireless World Wide Web referred to as W4 was built [6] and became among the first web clients that could function over wireless networks to access the Web. Digestor was an HTTP proxy that was published in 1997 and could transform pages suitable for a specific device [7].However it could only be configured for one user at a time and was tested mainly using PDAs. Dikaiakos [8] provided an overview of intermediaries which act as go-between for clients and servers. These include notifications and wireless-web proxies that are important components for pervasive web access to this day.

2 Papers' Summary We have chosen five recent research papers from the year 2000 to 2007 based on their relevancy with respect to web access in pervasive environments while also taking into account their number of citations as an indicator about their acceptability among researchers. We will now proceed with a summary of each paper in chronological order. More details about the selection process is available in Appendix A.

2.1 Paper 1 In his paper entitled “On proxy agents, mobility, and web access ” [9] in the year 2000, Anupam Joshi emphasised on the fact that most solution that researchers were coming with at that time for mobile access to the web involved proxies. He argued that proxy-based approaches are not always the best one specially for ad hoc networks where scalability can become a problem. End-end techniques without intermediaries between clients and server can be more appropriate in some of these situations. Therefore a hybrid approach was implemented involving a combination of both proxy-based and end-end approaches.

2.2 Paper 2 “Multibrowsing: Moving Web Content across Multiple Displays” [10] by Brad Johanson et al suggested that the missing link in making ubiquitous computing as successful as the Web is the lack of a framework that can enable browsing of information using multiple displays in both mobile and static environments. As a result, they proposed a framework named multibrowsing that extends the “information browsing metaphor of the Web across multiple displays”. This approach of tackling Web browsing is quite different from most others and therefore may be interesting to consider.

2.3 Paper 3 “Adaptive Interfaces for Ubiquitous Web Access” [11] highlights the fact that browsing the Web from mobile devices can become cumbersome and inefficient if no attempt is made at using adaptive personalisation technology to customise the information according to the users' needs and devices' capabilities. Therefore the solution proposed was one that could adapt itself based on users' preferences derived from a combination of similarity-based, collaborative and Bayesian methods.

2.4 Paper 4 In “A Web Browsing System based on Adaptive Presentation of Web Contents for Cellular Phones” [12], a scaleddown image version of a web page requested by the client is provided. Known as the overview, it is divided into a number of components which the user can select to zoom in and also read. Depending on the nature of the component selected, the latter will be presented accordingly, for example, where images will be zoomed out to fit the screen and can also provide auto-scrolling options users. The system was said to be effective but it was argued that due to lack of Web Access in Pervasive Environments: A Review

2/7

space, details of the testing has not been included. However this innovative browsing method can be quite interesting to think about.

2.5 Paper 5 “Enhancing pervasive Web accessibility with rule-based adaptation strategy” [13] aims at decreasing the time to access information required by a user in a mobile environment, the user's context which includes details such as the situation, kind of network, type of device being used and also about the user's disability status. Using these information,adaptation rules are derived. The proposed system then select the appropriate rules to come up with an adaptation strategy which will determine how to deal with each object in a web page. The transformed web page is then presented to the user. This approach has been proven to be efficient through lab experiment and also the adoption of a modular approach also means that it can easily be extended for future situations.

3 Critical and comparative evaluation “On proxy agents, mobility, and web access ” [9] makes a claim that the problems that were present some eight years ago mainly were about bandwidth and other resources such as memory and cpu being less in wireless mobile devices. These problems will continue even after apparently solving them since by then bandwidth and resources on static hosts will have also increased and applications developed to make use of them thus leaving the bandwidth and resource gap a continuing problem. Indeed to this day, the gap still exists and it seems it will continue. For [9], the major problem with having an intermediary between a clients and servers is that it does not scale well. Known as proxy-based approach, it is said that problems may arise when many clients request for pages that require CPU-intensive processing by the proxies. But eight years later, we find that processing power has increased tremendously and therefore the problem of processing bottleneck may have already been solved: it can be possible to have a cluster of proxies for example. In ad hoc networks, we also suggest the use of distributed computing whereby capable nodes can collaboratively act as proxies with an elected node coordinating the tasks among them. The approach adopted by the paper suggested the use of proxies and end-end whereby processing tasks are performed either by proxies or on the servers. At that time PDAs had little processing power but this is no longer the case, PDAs and smartphones have much more processing capability now. Therefore an alternative approach could also include some adaptation tasks performed locally on the devices themselves. This will enhance scalability since the burden on proxies and servers will be reduced. We therefore suggest that the new capabilities of mobile devices should be exploited to a maximum rather than treating them as dumb clients that should be fed with fully processed data. To choose which tasks can be done on the servers rather proxies, processing-intensive and also jobs that can be performed offline were chosen. For requests requiring on-the-fly transformations, these were left for proxies. But in case that most jobs begin to require this kind of transformations, then the whole idea of load balancing between proxy and server becomes irrelevant. This is perhaps the reason behind the non-popularity of this kind of hybrid approach. Experimental results were used to demonstrate the enhanced performance but these were performed in closed environments and therefore does not reflect reality. The use of multiple display units [10] is very interesting in the fact that it allows collaborative work. Most of the work in that paper focused on this aspect. However, a possibly new scenario is a user with a mobile device having small display is advised by the latter about the location of a larger public one in close-proximity where information can be displayed more appropriately for the user's own convenience rather than for a group of people working together. The research performed was not about transforming the data to adapt to devices but rather to enable its migration or reproduction from one device to another thus by-passing the need for adapting a webpage for a particular device. The testing phase was performed in a controlled environment, referred to as the “iRoom” which has the infrastructure for a ubiquitous conference including wall size display. Divided between clients and targets, the framework requires that clients have a Java-based plugin known as MB2GO to be installed while targets should run a service called butler implemented using ICrafer. The MB2GO plugin is browser dependent and is compatible with only Microsoft's IE. A major difference that this framework has, compared to others, is that it does not rely on any centralised proxy or server, instead, any client can 'pull' or 'push' webpages into target displays making it a very robust solution. [11] identifies the problem about sending all information from a webpage to a small-size screen on a mobile device causing an overload at the user's end and also bandwidth wastage over the network since much of the data sent will not be useful to the user. It was tested over a six-week period by comparing two groups of users accessing news using mobile devices; one group receiving personalised content and the other receiving normal content. The former group tended to return to the website more often. Tested for a news website only, the proposal yielded positive results although additional real experiments with other kinds of websites would have been desirable rather than just theoretically making

Web Access in Pervasive Environments: A Review

3/7

up scenarios as done in this paper. However it has been experimented with text-only mobile devices. Thus with the latest high-resolution screen mobile phones and differing display size, this system will lead to under-usage of the devices' capabilities specially since technologies keep improving almost exponentially. Presenting a proxy-based solution, [12] the paper suggested that mobile phone screens are not wide enough to accommodate whole web pages at one time thus making it difficult for users to navigate over different parts of the page without losing the context. The solution proposed to this problem is by breaking webpages into different components such as a list of links, tables of data and images and then allow users to view them individually by making a selection from a reduced scale of the page. Although extracting components and classifying them are crucial for the whole system to work, very little information is given about this step. However considerable experimentation was performed involving sixteen test-subjects to check the auto-scrolling feature and how the users normally read each component classes. While auto-scrolling can be an interesting feature for mobile phones with no button-only input capabilities, there is a trend of phones equipped with touch-screens. Whether auto-scrolling will still be attractive in these situations remains to be found out. The proxy server software component responsible for the adaptation task is coded in C# and PHP which is a rather unusual combination. But no justification is made about this choice nor about the phone used (SH900i model), equipped with i-appli developed by NTT Docomo, and coded with Java. According to the conclusion section, the solution still has problems in the classification of components task. This kind of problem with become even more complex in the future since more hybrid and perhaps new classes of components will be in use on web pages, leaving the efficiency of this system in uncertainty. [13] also try to tackle the problem of small size display in mobile devices when browsing the Web. It adopts a “modular design method” when developing adaptation rules that are then used to generate strategies through an expert system. This makes the system more maintainable and very extensible. Being a proxy-based solution, whenever a user makes a request, the proxy, referred to as the adaptation server get the user's context which is then used to generate the adaptation strategy that enables the selection of appropriate objects to make up the adapted content. These objects are obtained by parsing the page requested by the user, decomposing it into components that are stored as a tree structure. A number of scenarios have been tested and proven to work adequately with the solution. Most of the papers analysed so far, are proxy-based and leaves the security aspects of these solution untouched. However communicating through proxies can be vulnerable to man-in-the-middle attacks and thus unsafe for financial transactions for example. Selective end-end communication as suggested in [9] can be an option in this case and in addition to choosing only CPU-intensive or offline tasks, those requiring secure communications should also be endend. We must also highlight that with ubiquitous computing, it is expected that users are at the centre of the attention instead of devices. Thus context awareness is crucial in the designing of systems attempting to solve ubiquitous problems. While some papers take user-context as a factor in providing adapted contents, others simply take the only the devices' capabilities into account. Proxy-based solutions have been found not to scale up very well and can be a problem when devices are not aware about its address for example. Some kind of broadcasting mechanism will have to be used to discover them, probably the Web Service Dynamic Discovery (WS-D) [14] can be useful in this case.

4 Conclusion We have seen that although many solutions are being proposed, we have yet to achieve a reasonably proven, securityenabled, scalable and effective one for accessing the Web in ubiquitous and pervasive computing environment. The main reason is probably because the Web is in continuous evolution and thus further research have to be carried out to produce corresponding ubiquitous access methods. Nevertheless, we need solutions that can adapt to the constantly mutating Web so that the relationship between ubiquitous computing environments and the Web can thrive, without forgetting to bring the aspect of security on board.

5 References 1.

Mark Weiser, “The computer for the 21st century” Scientific American (1991)

2.

“WorldWideWeb: Proposal for a HyperText Project,” http://www.w3.org/Proposal.html. Accessed on 25 December 2008

3.

Stefan Gessler and Andreas Kotulla, “PDAs as Mobile WWW Browsers,” Computer Networks and ISDN Systems 28 (1994): 53--59, doi:10.1.1.48.9198.

Web Access in Pervasive Environments: A Review

4/7

4.

T. Watson, “Application design for wireless computing,” in Mobile Computing Systems and Applications, 1994. Proceedings., Workshop on, 1994, 91-94.

5.

G.M. Voelker and B.N. Bershad, “Mobisaic: an information system for a mobile wireless computing environment,” in Mobile Computing Systems and Applications, 1994. Proceedings., Workshop on, 1994, 185190, doi:10.1109/MCSA.1994.513481.

6.

J. F. Bartlett, “Experience with a wireless world wide web client,” in Proceedings of the 40th IEEE Computer Society International Conference (IEEE Computer Society, 1995), 154, http://portal.acm.org/citation.cfm? id=527213.793598.

7.

Timothy W. Bickmore and Bill N. Schilit, “Digestor: device-independent access to the World Wide Web,” Computer Networks and ISDN Systems 29, no. 8-13 (September 1997): 1075-1082, doi:10.1016/S01697552(97)00026-3.

8.

M. Dikaiakos, “Intermediaries for the World-Wide Web: overview and classification,” in Computers and Communications, 2002. Proceedings. ISCC 2002. Seventh International Symposium on, 2002, 231-236.

9.

Anupam Joshi, “On proxy agents, mobility, and web access,” Mobile Networks and Applications 5, no. 4 (December 1, 2000): 233-241, doi:10.1023/A:1019120915034.

10. Brad Johanson et al., “Multibrowsing: Moving Web Content across Multiple Displays,” in Ubicomp 2001: Ubiquitous Computing, 2001, 346-353, http://dx.doi.org/10.1007/3-540-45427-6_29. 11. Daniel Billsus et al., “Adaptive interfaces for ubiquitous web access,” Commun. ACM 45, no. 5 (2002): 34-38, doi:10.1145/506218.506240. 12. Yuki Arase et al., “A web browsing system based on adaptive presentation of web contents for cellular phones,” in Proceedings of the 2006 international cross-disciplinary workshop on Web accessibility (W4A): Building the mobile web: rediscovering accessibility? (Edinburgh, U.K.: ACM, 2006), 86-89, doi:10.1145/1133219.1133234 13. Stephen J. H. Yang and Norman W. Y. Shao, “Enhancing pervasive Web accessibility with rule-based adaptation strategy,” Expert Syst. Appl. 32, no. 4 (2007): 1154-1167. 14. “WS-Discovery Specification” http://specs.xmlsoap.org/ws/2005/04/discovery/ws-discovery.pdf. Accessed on December 12, 2008

6 Bibliography 15. M. Satyanarayanan, “Pervasive computing: vision and challenges,” Personal Communications, IEEE [see also IEEE Wireless Communications], vol. 8, 2001, pp. 10-17. 16. C. Patterson, R. Muntz, and C. Pancake, “Challenges in location-aware computing,” Pervasive Computing, IEEE, vol. 2, 2003, pp. 80-89. 17. K. El-Khatib, Z.E. Zhang, N. Hadibi, and G.V. Bochmann, “Personal and service mobility in ubiquitous computing environments,” Wireless Communications and Mobile Computing, vol. 4, 2004, pp. 595-607.

7 Appendices 7.1 Appendix A: Selection Process The original idea as about choosing a topic involving ubiquitous computing environments since this seems to be a very active research area and presents many challenges. Since this area is very vast, we have chosen to emphasise on “accessing services in ubiquitous environments” and started a literature review by searching for papers related to these terms in Google Scholar. Very soon it was found that in order to be able to explore an area better, more focussing is needed rather that the broad area of service access in ubiquitous computing environments. Since our field of study is Web Technology, it was thus decided to narrow our topic to “Web Access in Pervasive Environments”. The next step was to find papers related to this specific topic by searching with keywords such as: •

ubiquitous web access

Web Access in Pervasive Environments: A Review

5/7

•

pervasive web access

•

web access with mobile devices

•

web access in ad hoc networks

in Google Scholar, ACM and IEEE portals. This yielded hundreds of papers with the most relevant ones being in the first 3 pages. From these results, based on the titles, the ones that seemed interesting were added to a new list. Using the list, each of the items were followed up and the following criteria were taken into account to give each of them a score: relevancy of abstract to our topic and publication date. This effective weeded out most papers that had misleading titles and those having been published more than 10 years before. Nevertheless, a few papers although outdated but which seem very relevant were kept to be used in the historical overview. The items from the list were skim-read and interesting references were followed up. This brought a few other interesting and relevant papers to our list which now consisted of 14 articles.

Since only five main articles are required, a more thorough selection process was performed involving additional criteria as detailed in the table below:

Blue colour means that the paper is among the best ones from our list. Orange ones are in the second level making them interesting but less than the blue ones. Finally the yellow highlighted ones are from the least relevant or with less acceptability among researchers. This is how the five main papers were selected for this report.

7.2 Appendix B. How this report influenced my ideas Before working on this report, whenever I heard about accessing the Web from other devices, specially smaller, mobile ones, my idea was about creating other versions for each websites by the original web-designers themselves. Now I know that although this is a possible way, it is very impracticable. The options that are more convenient are those that automate this conversion process, usually on-the-fly to adapt to a range of devices. Another approach can simply be to suggest the user to move to a nearby public display unit to view a website in some cases. However, there is considerable work to be done in the security area for these solutions and most papers have put security as a future work. Since security is crucial when dealing with publicly accessible systems, if we want these solutions to succeed, security should also be considered in parallel.

7.3 Appendix C. Lessons learnt When attempting to solve a problem, even in research, the divide and conquer approach works best. That is, big problems should be broken into smaller ones and they become simpler to solve. Number of citations is not always a good way of measure for a research paper in mainly two cases: Firstly when the Web Access in Pervasive Environments: A Review

6/7

research paper has been recently published and the second case is when a research paper tackles a problem area which rarely looked at by other researchers. Also, a heavily cited paper in no way guarantees that it is a good one since the references to it may be of negative nature. In this aspect, I would have liked to suggest that citation counts be broken into 3 categories as detailed below: + Citations: for references praising a paper ~ Citations: for neutral references using definitions or explanations or methodology of a paper

- Citations: for references of negative nature highlighting the drawbacks of a paper for example This will give a better idea to researchers when using citation count as a measure. We also found that many research papers propose solutions that are not tested in real-life scenarios and therefore cannot be relied upon when developing other solutions based on those.

Web Access in Pervasive Environments: A Review

7/7