Mobile Agents for e-commerce Rahul Jha Under the guidance of Prof. Sridhar Iyer KR School of Information Technology , IIT Bombay Overview Mobile Agent applications in e-commerce Mobility Patterns and implementation strategies Quantitative performance evaluation of Voyager Evaluation of Voyager, Aglets and Concordia Our Prototype of e-commerce application using mobile agents 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay e-commerce applications Involve – – – Characterized by – – Product search Order Placement and confirmations Negotiations Large amount of data exchange Client specific services Require – – Real time interactions Disconnected (or low bandwidth) shopping 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Mobile Agent advantages Mobile agents (MA) – “A mobile agent is a program that can autonomously migrate between the various nodes of a network and perform computations on behalf of the user” MA advantages – – – – – reduce network usage faster response times add client-specified functionality to servers increase asynchrony between clients and servers introduce concurrency 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Mobility patterns and Implementation strategies 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Implementation strategies 2 3 C 2 1 4 3 5 6 Client 4 1 Server C C (a) Sequential Client Server (b) Sequential Mobile Agent Mobile Agent Message exchange 123456 2 2 1 2 1 2 1 2 1 (c) Parallel Client Server 16th January 2001 2 1 1 C Numbers along the arrows indicate the sequence of messages./ MA movement. C (d) Parallel Mobile Agent M.Tech Presentation KReSIT, IIT Bombay Mobility Pattern Parameters Definitions Itinerary the set of sites that an MA has to visit static dynamic Order the order in which an MA visits sites in its itinerary. static dynamic 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Static Itinerary Static Order H1 H2 H3 H4 H1 H2 H3 H4 Order Itinerary C H1 H2 H3 Applicable Implementation Strategies 16th January 2001 M.Tech Presentation H4 • • • • Sequential CS Sequential MA Parallel CS Parallel MA KReSIT, IIT Bombay Static Itinerary Dynamic Order H1 H2 H3 H4 H1 Order Itinerary ? C H1 H2 H3 Applicable Implementation Strategies 16th January 2001 M.Tech Presentation H4 • • • • Sequential CS Sequential MA Parallel CS Parallel MA KReSIT, IIT Bombay H1 Itinerary Dynamic Itinerary H1 ? Order ? C H1 H2 H3 Applicable Implementation Strategies 16th January 2001 M.Tech Presentation H4 • • • • Sequential CS Sequential MA Parallel CS Parallel MA KReSIT, IIT Bombay Experimentation and results The e-commerce application – – A single client searching for information about a particular product from the catalog of several on-line stores We assume that the client requires a highly customized search which the on-line store does not support. 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Experimentation Experimental setup – – – Voyager™ Framework for MA implementations Java™ socket based implementation for client server interaction On Pentium-III, 450 MHz workstations connected through a 10 Mbps LAN with typical student load 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Parameters Parameters Range number of stores size of catalog size of client-server messages processing time for servicing each request network latencies on different links 1 to 26 20 KB to 1 MB ~ catalog size 16th January 2001 10 ms to 1000 ms assumed constant (all workstations on the same LAN) M.Tech Presentation KReSIT, IIT Bombay Performance metric User Turnaround Time time elapsed between – a user initiating a request and receiving the results. equals time taken for agent creation + visit / collect catalogs + processing time to extract information. 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Effect of catalog size on Turnaround Time 14 Turnaround time (sec) 12 10 MA 8 CS of catalog size 100K 6 CS of catalog size 200k CS of catalog size 500K 4 CS of catalog size 1MB 2 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 No. of shops visited 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay processing = 20ms 16 Turn around time (sec) 14 12 Sequential MA 10 8 Parallel MA 6 Sequential CS 4 Parallel CS 2 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 No. of shops visited 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay processing = 500ms 30 Turnaround time (sec) 25 20 Sequential MA 15 Parallel MA Sequential CS 10 Parallel CS 5 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 No. of shops visited 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay processing = 1000ms 45 40 Turn around time (sec) 35 30 Sequential MA 25 Parallell MA 20 Sequential CS 15 Parallel CS 10 5 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 No. of shops visited 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Observations Mobility patterns determine the implementation strategies Sequential CS most suitable where – a small amount of information has to be retrieved from few remote information sources. Parallel implementations effective when – processing information contributes significantly to the turnaround time. 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Observations Mobile agents out perform traditional approaches when – When the cost of shipping MAs < message exchange size. MAs scale effectively across the parameters of E-commerce application 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Evaluation of Voyager, Aglets and Concordia 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Qualitative Comparison Voyager Aglets Concordia ORB MA based framework MA based framework Java messaging Transparent No No Multicast Yes No No Publish/Subscribe Yes No No Scalability Space No No Authentication and security Strong implementation Weak implementation Strong implementation Agent persistence Yes No Yes Naming service Federated No No Remote agent creation Yes No No Grouping / Collective Logical Garbage collection Yes Features Category 16th January 2001 Physical No M.Tech Presentation Physical No KReSIT, IIT Bombay Quantitative Evaluation Experiments Mobility pattern : Product discovery Parameters Range number of stores 1 to 26 size of catalog 1 MB Message packet size Kept constant for all 3 frameworks processing time for servicing each request 20 ms network latencies on different links assumed constant (all workstations on the same LAN) Experimental setup : Same as that for previous experiments. Performance metric : User turnaround time 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Cost of message exchange User turm around time in sec Message exchange cost for different mobile agent frameworks 25 20 Concordia 15 Voyager 10 Aglets 5 0 1 2 10 50 100 200 300 400 500 Number of message packets Numner 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Cost of code shipment Code shipment cost for different mobile agent framework 2500 Time in ms 2000 Concordia 1500 Aglets 1000 Voyager 500 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 No of shops 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Observations Voyager supports almost the super set of functionalities and features as compared to Aglets and Concordia. Voyager being an ORB has advanced messaging support and hence performs much better than Aglets and Concordia. Cost of code shipment for Voyager is more than Concordia (both user RMI) – Voyager is an ORB with mobility support – Large set of functionalities supported by Voyager 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Our Prototype of e-commerce application using mobile agents 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Architecture of Our Prototype Model Buyer Buyer's agent Buyers GUI Product Request Template as XML List of shops to visit and dockyards SHOP SHOP SHOP Shopkeepers GUI Shops agent Sales Transaction Log 16th January 2001 DB DB M.Tech Presentation Salesman agent Salesman agent Salesman agent Local services Product Catalog KReSIT, IIT Bombay Interaction among Components Filtered Result 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Conclusion Helps user with tedious repetitive job and time consuming activities. Faster and real time interacting at shops Reducing network load Support for disconnected operation. Introduce concurrency of operations Client specific functionalities at the shops 16th January 2001 M.Tech Presentation KReSIT, IIT Bombay Thank You