Paper 10

Paper Title:User-Centric Services Provisioning in Wireless Environments

Quan Z. Sheng, Boualem Benatallah, Zakaria Maamar
November 2008
Communications of the ACM , Volume 51 Issue 11

Three Critical Questions:


Group 1:

Member Name: Chiranjeevi Ashok Puvvula

1. Author says that the system or agents should be “Context aware”, so that they can better understand the situations in which the User is. From a semantic web point of view, this means that the agents should be intelligent enough to make suitable assumptions about the user needs. Does this mean that the agents take the control from the user hands based on the user preferences, user locations? This can’t be done always because users want to make their own decisions?
2. The use of XML based languages for communication or discovering services has some problems. Even though XML is known for its simplicity, it places a huge burden on the client system. This is because of the parsing to be done. This parsing needs lots of CPU resources and memory to do all the comparisons (i.e. the processing power). In addition to this, these XML documents need to travel between the agents in a high through put networks which requires more bandwidth.
3. As said, different companies use different protocols for message communication taking into account reliable data transfer and security. Communication between users(customers of different companies) requires building new protocols each time. This would be very complex. How can this issue be resolved?
4. The architecture proposed in the system requires lots of communication between the layers and in between the agents. There is a chance of “high run time overhead”. How can this issue be resolved?
5. How can the agents specified in the architecture be fully automated in doing their activities?

Group 2:

Member Name:Sai ram Kota

1>“Like markets for goods and services, the knowledge market has buyers, sellers, and brokers, as well as market pricing and exchange mechanisms, even though money is rarely the form of payment “. What do you think are other modes of payment for an ordinary business company..??
2>Authors say “In electronic knowledge trading one cannot simply copy ways of working that are already known from traditional business, but should exploit the strength of multiple synchronous and asynchronous communication..??” what are these synchronous/asynchronous communications. How are they fundamentally different from traditional approaches..?
3>The authors generalize that even services such as manufacturing can benefit by exposing their knowledge assets..?? But for a manufacturing service most of its data is confidential. (Amount procured, quantity manufactured, costing mechanism). We did not know what and how exactly can they benefit by opening their valuable knowledge assets especially when it is open to competitors..??
4>Authors say that “We can’t replicate e-commerce between participants who share”. But when a user will have to pay for a offered web based knowledge service, you tend to do a do commerce on web. So how is it different from traditional e-commerce?

Group 3:

Member Name: Sunil Kumar Garrepally

1. Usage of XML based web services makes the architecture platform, device and vendor independent. But, representing data in XML format requires additional parsing of XML data that would require more battery power. So, would this parsing suitable for mobile devices that are a major requirement for the mobile applications?
2. The author has proposed an architecture that contains 4 layers. But, the author has forgotten to include business layer which allows transaction management or helps to define the business processes. Is this architecture feasible for real time applications without including the business layer in it?
3. The events are explained clearly using sequence diagram. But, it shows that there are several messages via several agents involved for composing a single service. In case of composing the complex services, this process becomes more complicated and inefficient. Is this architecture suitable for large scale environments that need multiple EAs and CAs have to be deployed?
4. The main problem with the mobile devices is the connection loss because of network problems. To, avoid this it is very important to pause the session instead of terminating it, and continue when the device is connected to the network. Does this architecture also support session concept that helps to maintain the sessions and retrieve sessions in case of connection failure?
5. The proposed architecture contains registry that is used for searching corresponding services. What does it do when the required service is not found in the registry? Does it throws any error message or informs the service provider for the requirements of user?

Group 4:

Member Name:Prashant Sunkari

1. The paper stated that the proposed architecture is well-suited for “small domains” only. If this architecture is to be extended for the large domains then additional agents and features are to be added. This increases the complexity and overhead. How this architecture is suited for large domains reducing the complexity is not explained in the paper.
2. The architecture proposed has different layers where data are to be exchanged between different layers. How the key aspects like security, reliability is achieved.
3. The author stated that the architecture supports only “asynchronous communication”. But to enhance “throughput, performance and reliability”, we need “synchronous communication”. Why this architecture does not support “synchronous communication”?
4. The paper stated that the “major challenge” for the “mobile devices” is the availability of “limited resources”. But how this architecture is used to eradicate this problem and how the services are deployed to meet the requirements of “mobile devices” is not stated in the paper.

Group 5:

Member Name: Ritesh Mengji

1)In the Design Principles/Agents section, the author says that the combination of services and agents will provide a self-managing infrastructure. What are the techniques used by the agents to build a self-managing infrastructure and how can they be fully automated?
2)The validation of the design was confined to a small set of people (18). The critical question is can we rely on the outcome of a small set of people against entire population of potential users of the system?
3)The authors proposed a layered architecture which sets certain limits to each layer. Does this limit the scalability of the system?

Group 6:

1) Is it very safe to depend completely on the decisions of the agents without taking user decisions into consideration?
2) Will the service agents provide all those features which the web services provide to the client? If they don’t provide some features, how can they be provided?
3) Where do we have the access point? Is it is between one user and all clients or between each and every user and a client because it is waste of having the access point between each and every agent and a client?

Group 7:

Member Name: Priyanka Koneru

1. “ There are few limitations in the wireless environments such as limited resources, low throughput , poor connectivity etc which make current web service provisioning techniques inappropriate. Will the ‘multi-agent based architecture’ described by the author in the paper overcome these limitations? Will the task of developing complex applications composing other services becomes easier using this architecture?”
2. “ The ‘multi-agent based architecture’ described in the paper provides asynchronous communication which decouples senders and receivers in both space and time which are characterized with frequent intermittent connectivity “. “ The author didn’t mention why he preferred asynchronous communication over synchronous communication? What are the disadvantages in using synchronous communication?”
3. “ The ‘ multi-agent based architecture’ described in the paper has four layers and it supports asynchronous communication. So how far this architecture ensures secured communication? Will this architecture support both synchronous and asynchronous communications ?“
4. “ The architecture mentioned in the paper has certain limitation that it supports only small domains. So what are the mechanisms for managing contexts and events among the agents to make this architecture support large scale domains”.
5. “ Can the architecture mentioned in the paper be extended to support large scale environments? What applications are to be built on the top layers and how is the priority given? How the large scale domains effect the performance of the architecture?”

Group 8:

Member Name:Hema Snigdha.P, Lakshmi Bhavani.N

1) The given example in this approach is about “class-assistant process” ,in which simplified state chart diagrams are quite complex then how about the entire state chart for the given example and for complex business application?
2) The approach which is mentioned in this article to implement mobile applications, was it implemented and was it beneficiary?
3) It is mentioned that there is only one “service agent” per service and is downloaded along with the tuple space. Can it not be re-used instead of using a new service agent for each service requested?


Group 1:

Member Name:pelluri,lattupalli,voruganti

Why doesn’t the author frame the architecture with the help of components ,the wireless environment is more a component natured environment than a agent oriented,so why does he use agents instead of components?

The author in the related work says that the issues addressed in the area of the semantic web are complimented to theirs,but then how does he relate this architecture to the issues of the semantic web,though it’s a service oriented approach,how are issues over interoperability and integration being explained in the architecture?

Event condition action,this is the usual format of the tuples(orchestration)how much data can be configured based upon this format,what if certain data is not related to a format ,what would be the consequences,Are tuples reliable enough for huge domains?

Group 2:

Member Name: Addagalla Satyanaga Manoj Kumar; Bobbili, Shrinath; Gopinath, Sreejith

 The system proposed by the authors assumes certain capabilities in all mobile devices. However, the critical question here is whether the system is flexible enough to allow devices with diverse capabilities, what contingency plans are present in the system design that plans to handle unsupported devices, and what features are required in the devices themselves that enable them to fully exploit features of the new system?
 The system proposes to use templates to define input processes; wont such a design decision necessarily restrict the capabilities of high-end devices to exploit system features? Does it not mean that that users are restricted to as small a subset of features as the system designers care to expose?
 A layered approach most often forces system designers and administrators to define the amount of system resources earmarked for each layer; the critical question is whether this limits the scalability of the system.
 The authors introduce a concept of tuples to trigger and fire operations. Along with the concept of such data interactions comes the problem of large-scale data management; the critical question here is whether the authors have thought of and provided for an online database that is fast and quickly indexable and searchable.
 In the evaluation section, the authors mention that the system has been tested on a small set of 18 users. The critical question is whether 18 users is a large enough group to be considered representative of the entire population of potential users of the system.

Group 3:

Member Name: Swathi M Shastry

1. Does the web service provision to mobile users provide ontological services? Can the semantic web feature be implemented in the mobile applications? It looks highly unlikely presently as the current system caters to only a small domain, but if semantic web services were provided, what additional functions would have to be implemented on the deployment and context agents?

2. In the event of an exception handling event occurring like the mobile application getting disconnected due to bad network or the battery dying down, the result gets sent to a transformation service as stated. What technique does the transformation service employ to keep a record of which mobile device to send which result to? Also there might be multiple mobile users in the same vicinity at the same time asking the same kind of queries, so if the network goes down, all of the results get sent to transformation services, so would they be sent to different transformation services or a single transformation service? How does the transformation service decide which result to send to which mobile user?

3. Is there any protocol to be followed by the deployment agent while assigning the transformation services at the time of orchestration tuples? Are the orchestration tuples chosen based on the user preferences or vicinity?

Group 4:

Member Name:Shaiv, karuna priya rameshwaram, anusha vunnam

1) Up to how far the proposed system will help to overcome the two challenges such as of providing personalized access and in extending the mobile resources?
2) Would it be beneficial to develop such a complex system for even small applications?
3) Even though the outcome of the proposed system had given good results for the small applications, would it be same for the large applications and what about the cost of developing such a complex system?

Group 5:

Member Name: Rahul Mootha,Rahul Reddy

• The author states that agents can perform active actions in dynamic environment, how are they done and how is it different from the architecture in wired environment?
• Can templates be requested directly or be defined by the user and provide it to the user agent which in turn is returned to the service layer?
• How are compostite and integrated services provided by the agents ? Is there any difference in how it is done in a normal environment to provide results to the user in a quicker way as it is the major limitation in mobile environment?

Group 6:

Member Name:

Group 7:

Member Name: Goyal, Saurabh, De Morais Andrade,Pablo and Boda,Vamshidhar Reddy

 If user mobile device is connected to secured and trusted network like campus or offices, it is secure to use the provided services. But if user uses under untrusted network like public network, will this system design able to provides the privacy and reliability?
 Author said that they used secure and reliable transmission but there is no explanation given how to provide security with so less computing resources at mobile device.
 How to provide security in heterogeneous environment or foreign environment? Any user can take other user data or work without their permission.

Group 8:


1. In the paper the author mentioned several agents in the system design. Also he mentioned that the agents will use the policies and their own knowledge for taking actions. Also the agents can be able to take decisions in the dynamic environments. Is it safe to rely completely on the agents without taking the decisions of the user?
2. The author mentioned that the client is an application which will be run on the mobile devices. Also there is access point between the user agent and the client. Is the access is between every user agent and the client or between the one user agent and all the clients?? Because if it between one user agent and one client then it is waste to have the access point between them.
3. In the paper the author mentioned that the service agents will act are the proxies for the web services which the client had requested and for this the service agent will require only the orchestration tuples. IS it possible that the service provides all the features that the web services will provide for the client? If not how the other issues can be solved?

Group 9:

Member Name: Satish Bhat, Holly Vo

1. How much storage does the new architecture reside on mobile device? What are the minimum requirements for a device to apply this architecture?
2. How does the system act as a passive listener (comparing to traditional system)?
3. What types of mobile application/activities can appropriately act on this architecture?

Group 10:

Member Name: Sunae Shin, Hyungbae Park

1) There is a lack of explanation for making decision of protocols laid foundations for the profile of user agent. For instance, how can personal preference be the source to deciding interaction protocol? The specified process from profile information to results needs to be examined.
2) The more technical statements and examples will make clearer to understand. The only one example, battery, is mentioned to describe events for orchestration interactions.
3) Assumptions may need for participants’ of evaluation. During the evaluation, 18 people are participated and they overviewed the system architecture. They might not understand or quite interested in the system. Thus, information of participants’ need to be specified such as where they come from, what is their interesting area or they are picked randomly. This may effect to the evaluation.
4) The questions that they asked to participants’ are limited. The questions about level of collected data or specified amount of receiving data need to be shown.
5) The system can be more efficient if the communication between agents can be controlled. When communication is failed the system can be overloaded or make confusion

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License