Abstract—

With increasing use of the E-commerce for the business over the cyber space, security threats are also increasing. The trustworthiness of the business over the internet has become critical both for the consumer as well as for the service provider. In this paper I’m going to discuss how Intelligent Agents can help to securely exchange the information while doing a transaction over the internet. I will evaluate the current techniques used by the multiple agents, especially for the agents developed for the E-commerce. Also, analysis of the agent based system in regards of their ability for tackling threats.

INTRODUCTION:

There are numerous threats to the security of Internet Ecommerce. Security breaches are most frequently discussed in terms of the information as it can modify the content of the message. The internet is only one potential source of insecurity; further elements of the problem are:  The Customer side where a customer can be impersonated, with or without the use of the customer‟s equipment. The use of stolen credit card details is the simplest example.  The vendor side where the vendor can trade inappropriately or dishonestly 

Intelligent Agents & E-Commerce

E-commerce has a lot of advantages such as ease in accessibility, low operating cost and broader services. But, there are some barriers blocking the road to success, which include overload of information, difficulty in searching, lack of negotiation infrastructure, etc. This is why e-commerce need advanced technologies as support. Intelligent agents are seemed to be the excellent candidate with their properties of intelligence, autonomy and mobility. Intelligent agents act on the behalf of the customers to carry out delegated tasks automatically. They have demonstrated tremendous potential in conducting various tasks in e-commerce, such as comparison shopping, negotiation, payment etc

Intelligent Agents & E-Commerce E-commerce has a lot of advantages such as ease in accessibility, low operating cost and broader services. But, there are some barriers blocking the road to success, which include overload of information, difficulty in searching, lack of negotiation infrastructure, etc. This is why e-commerce need advanced technologies as support. Intelligent agents are seemed to be the excellent candidate with their properties of intelligence, autonomy and mobility. Intelligent agents act on the behalf of the customers to carry out delegated tasks automatically. They have demonstrated tremendous potential in conducting various tasks in e-commerce, such as comparison shopping, negotiation, payment etc. .

B. Intelligent Agents Security & Trust Intelligent agents are the programming paradigms which allow flexible structuring of distribution of the computation over the internet. But, before the agents can be used in the ecommerce applications their security must be defined. Most of the intelligent agent systems refer to four elements for their security, which are as following:

 A secure runtime environment (e.g. the Java Virtual Machine) for host protection

 Code signature to prove that the agent has not been tampered with

 Host authentication to prove that the agent is about to move to the intended host

 A secure channel over which agent can migrate.

C. Different stages in Working of MAS The working of MAS varies from Model to Model of different MAS. But there are few aspects of working of MAS which are almost common in an e-commerce transaction. According to Singlee and Preneel (2004) there are Different stages in the scenario, which are as following:

1) User sends agents: User generally connects to the internet periodically for a short time. As per user‟s requirements he sends multiple customized agents on the internet for the best bargains of different needs such as hotel booking, air ticket reservation and car rental etc. Thus, multiple agents have to communicate each other to fulfil the overall requirements of the user.

2) Agents travel and securely collect data: The agents travel from one platform to other. Only the data collected by the agents and their parameters are securely transferred between the agent platforms, but the program is accessed at its original location. The agents collect the information from each agent platform (Singlee and Preneel, 2004).

3) Agents conduct a secure payment transaction: An agent at any time can decide about conducting a financial transaction. Thus, it communicates along with other agents to generate the digital signature. All the agents together can generate a new signature for a particular offer at an agent platform. A combining entity is required to combine an appropriate set of the contributions of the agents into the resulting signature (Singlee and Preneel, 2004).

4) Verify collected data: In the end the agent returns to the semi trusted platform. This platform verifies for the authenticity of the collected data. That data includes the various transactions conducted by this agent.

various types of agents are shown which are as following:

1) Supervisor Agent: A supervisor agent works as help to the human supervisor for assigning the initial information, monitoring the access to confidential information and controlling the work flow for access to confidential information.

2) User Interface Agent: A human user can interact only with the user interface agent. Human user requests for the confidential information with the help of user interface agent. This agent works with authorization agent and allows the user to access authorized information.

3) Authorization Agent: This agent I responsible for that whether request for the confidential information is valid or not (Seo et al., 2004). E. Architecture of Multi-Agent Systems (MAS) Architecture of the Multi-Agent Systems depends upon the need of the system.

There are various models of the MAS as different models are developed for the different systems working. According to Marik and McFarlane (2005), Agent architectures usually have two recognizable parts:

 The Wrapper accounts for inter-agent communication and just-in-time reactivity.

 The body carries out the agent‟s main functionality. Several typical agent architectures exist, ranging from purely reactive agents, which operate in a simple “stimulus-response” fashion, to more deliberative or goal oriented agents, such as belief-desire-intention agents, which proactively reason about their goals and actions (Marik and McFarlane, 2005). Various Models of Agent based systems are there to fulfil the needs of a particular system, which differ in their architecture and working.

Such as:  MAGNET (Multi AGent NEgotiation Testbed): A Multi-Agent System for doing auctions on the internet.

 RESTINA: It a reusable multi-agent infrastructure for building agents with capabilities of inter-agent, message-passing communication. The RESTINA infrastructure can be used by agent developers to quickly develop agents for different applications. RESTINA was developed having Internet applications in mind (Wong and Sycara, 1999)

 SAFER: SAFER is an infrastructure designed to serve agents in e-commerce and to establish necessary mechanisms to manipulate them. It focuses on three fundamental activities of agents, namely, fabrication, evolution and roaming (Guan et al., 2002). In the following section of this paper, MAGNET is going to be discussed, such as its Architecture, Security drawbacks and Trust assumptions etc. to evaluate it. 

SUMMARY & CONCLUSION

After investigating the Intelligent Agents for Ecommerce, I have come to this point that, however ecommerce is coming up as a big business there are few barriers in the success of that because of Trust and Security. With the help of Intelligent Agents a high level of security can be implemented and trust can be gained. But, it is very difficult to make such intelligent agents which are perfect can be fully trusted. Also it will take time before people can trust agents and use them.