This page is about the research I have done while working towards my Ph.D. degree at University of California, Santa Barbara. The objective of my Ph.D. research has been to design and implement an e-commerce system for comparison shopping over the Internet using mobile agents and to study the market economics that have evolved in digital markets.
Disseration Title:
Mobile Software Agent Enabled E-commerce: System Design and Profit Maximizing
Algorithms.
Table of Contents
Ph.D. Thesis in compressed postscript format thesis.ps.gz
(2.12MB)
Ph.D. Thesis in PDF format thesis.pdf (2.77
MB)
During the first year of my Ph.D. research, I developed the Mobile Agents for Networked Electronic Trading(MAgNET) system. MAgNET uses Java mobile software agents called Aglets to comparison shop for an item from different sellers. Comparison shopping by buyers utilizes the pull model of marketing where the items sold by online sellers have sufficient value to attract customers. MAgNET can also be used for the push model of marketing where the sellers approach buyers proactively to sell their items. Reduced time for inventory clearance, rapid marketing of the product to a large number of buyers and dynamic price determination makes MAgNET attractive for implementing the push model. The software and tools used in MAgNET are Java 1.1.7, Java Swing SDK, IBM Aglets SDK 1.1, AELFRED XML Parser, Mini SQL 2.0.4).
In traditional markets, sellers compete with each other to maximize their profits. The use of automated comparison shopping aglets in MAgNET escalates the rapidity with which online sellers compete with each other. Thus, it is necessary for a seller to revise its price rapidly and accurately so that it can continue to obtain the maximum possible profit. We have developed the model optimizer algorithm that enables a seller to dynamically determine the optimum price that maximizes its profits under the present market conditions without explicit knowledge about its competitors' prices and profits. The algorithm then tries to increase the profit by dividing the buyer population into price-sensitive and price-insensitive segments and charging a different price to each segment.
The profits to the buyer and seller can further be increased by exposing concealed profits. In conventional markets, buyers and sellers usually conceal their internal profits. We have analyzed a scenario where buyers and sellers negotiate a deal after exposing their hidden profits and shown that the profits, both for the buyers and the sellers, are improved by this strategy. We have also shown that exposing hidden profits is the dominant strategy when the market comprises multiple buyers and sellers, and is therefore attractive to buyers and sellers.
Finally, we have analyzed the security and reliability aspects in MAgNET. Mobile agent systems are exposed to the malicious agent problem and the malicious host problem. We have developed a PKI(public key infrastructure) based security model that authenticates the buyer and seller's identity before allowing the mobile shopping agent to execute. We have also developed a fault-tolerance protocol over the security mechanism that makes the mobile shopping aglet resilient to network and node failures.