USING MOBILE AGENTS TO SUPPORT M-COMMERCE

Xining Li

Department of Computing and Information Science, University of Guelph, Guelph, Canada

Keywords: Mobile Device, Mobile Agents, M-commerce, Service Discovery, Database Service.

Abstract: Mobile agents are self-contained processes, dispatched by their principal, roaming the Internet to access

data and services, and carrying out their decision-making and problem-solving tasks remotely. Mobile

commerce (M-commerce), the traditional E-commerce combined with mobile devices and wireless

networks, is likely to become a major business model in the near future. Due to the flexibility and mobility,

mobile agent based M-commerce can complement the existing client/server based E-commerce model to

enable consumers to conduct business without time and space restrictions. For the purpose of applying

mobile agents to various internet applications, we have implemented an experimental mobile agent system

infrastructure that provides programming languages and virtual machines, and also integrates security,

service discovery and database access functionalities. The goal of this research is to deploy the existing

system infrastructure for M-commerce applications. We hope that the proposed research will benefit mobile

consumers who wish to access a wide choice of products and services on an anywhere and anytime basis.

1 INTRODUCTION

E-commerce is a business model of buying and

selling of goods and services via the Internet. This

model typically involves economic activities,

interactions between consumers and producers, and

commercial transactions crossing companies. There

is no doubt that the E-commerce is changing

economy and having a great influence on people’s

daily life. For consumers, E-commerce makes it

easier and more efficient to search, evaluate, and

compare products in the world market. For business

organizations, E-commerce can be used to raise

profit by increasing revenue and decreasing cost,

and explore new opportunities and expand business

into global market.

Most E-commerce applications use traditional

client/server model in which a commercial

transaction generally requires a stable

communication connection being established

between the client and the server. In recent years,

technological evolution has let to handheld

computing, such as PDA’s, mobile phones, Pocket

PC’s, etc., with combinations of wireless networks

including WiFi, Bluetooth, infrared, and GPRS or

3G telecommunication techniques. According to

IDC, almost 20 million units were shipped

worldwide in the second quarter of 2006, a 42.1%

increase, year over year. Due to technological

advances, a new E-commerce model, namely, M-

commerce has emerged and attracted a growing

number of research efforts (Shi, 2004, Bădică,

Ganzha and Paprzycki, 2005, Bai, Chou,Yen and

Lin, 2005, Kowalczyk, Braun, Frankczyk and Speck,

2003). In general, M-commerce can be identified as

the transaction conducted through the use of mobile

handheld devices over wireless or

telecommunication networks. M-commerce not only

extends Internet-based E-commerce, but also offers

a unique business opportunity with its own features,

such as ubiquity, accessibility, portability, etc.

Obviously, traditional client/server approach poses a

barrier to the development of M-commerce

applications. First, it will become expensive and

unreliable when lot of data has to be transferred

between the client and the server. Second, it will be

impossible to retain long time connectivity between

the client with a mobile device and the server. Third,

it typically requires clients to frequently check

trading opportunities and make most decisions

manually. In addition, compare with the desktop

oriented client/server or browser/web-server model,

mobile hand-operated devices have some physical

constraints, such as small screen size, poor network

connectivity, low transmission bandwidth, and

limited battery capacity. Therefore, in order to ease

the access and participation of mobile users, the

mobile agent paradigm has been increasingly

recognized as a promising framework for developing

Li X. (2009).

USING MOBILE AGENTS TO SUPPORT M-COMMERCE.

In Proceedings of the International Conference on e-Business, pages 91-96

DOI: 10.5220/0002189500910096

 SciTePress

M-commerce applications. Mobile agents are self-

contained entities capable of autonomously roaming

the Internet and launching user assigned tasks. This

paradigm brings forward the creative idea of moving

user defined computations towards network

resources, and provides a whole new architecture for

designing M-commerce applications. Deploying

mobile agents in M-commerce can reduce

unnecessary network traffic, tolerate poor network

connectivity, provide more advanced services,

support automation of decision-making, reduce

participation costs and improve trading efficiencies.

In this paper we propose an M-commerce

framework based on the Intelligent Mobile Agent

Gliding On-line (IMAGO) system (Li, 2006). The

IMAGO system is an infrastructure for developing

mobile agent based applications, such as M-

commerce or distributed data mining. In other

words, it provides an algorithmically complete

programming language and execution environment.

Based on the IMAGO system infrastructure, the goal

of this research is to use mobile agents to support M-

commerce applications. We hope that the proposed

research will benefit mobile consumers who wish to

access a wide choice of products and services on an

anywhere and anytime basis. The remainder of the

paper is organized as follows. Section 2 gives an

overview of the IMAGO M-commerce framework,

and describes the role of mobile agents on trade

transactions. In Section 3, we introduce three

important system facilities, namely, service

discovery, database management and agent

communication, and discuss their usage and

functionality. Section 4 presents the IMAGO

security mechanism. In Section 5, we present the

design of Mobile Portal that bridges a mobile user

and an M-commerce application, and show a simple

experimental example. Finally, we give the

concluding remarks as well as future work.

2 OVERVIEW OF THE IMAGO

M-COMMERCE FRAMEWORK

There is a great range of M-commerce applications

(Ngai and Gunasekaran, 2007). Typical applications

include mobile advertising, mobile inventory

management, product locating and searching, mobile

shopping, mobile entertainment services, location-

aware services, mobile financial applications, and so

on. Accordingly, there are many possible business

scenarios for developing M-commerce applications.

At this stage, our proposal focuses on the consumer-

driven M-commerce, i.e., an E-commerce model in

which the consumer initiates a trade transaction.

More precisely, our framework represents a

distributed environment that allows consumers to

dispatch mobile agents from their handheld devices

to visit E-stores for searching, comparing,

evaluation, buying and payment of goods. Figure 1

gives an example of agents occurring in the IMAGO

M-commerce system and indicates their basic

behaviours.

call back

IMAGO service

discovery server

Home IMAGO server

Figure 1 An Example of Deploying Mobile Agents in M-commerce

Remote IMAGO server A

MA1

vendor

agent

MA2

Remote IMAGO server B

MA1

vendor

agent

MA3

home

agent

MA1

DBM S

migrate

clone

communic at e ACL

DBM S

create

user

consumer

PDA

dispatch

locating

agent

Figure 1: An example of Deploying Mobile Agents in M-

commerce.

Generally speaking, the IMAGO M-commerce

framework distinguishes three major types of agents:

• Device agent: It is bound with user and

installed on the handheld device. Its major

responsibility is to enable a mobile user to

locate its home server, communicate with

the home agent to invoke an M-commerce

application. We call the device agent as

IMAGO Mobile Portal.

• Stationary agent: A stationary agent always

resides at its host. For different purposes,

we identify three kinds of stationary agents:

home agent, locating agent and vendor

agent. Home agent provides a bridge

between the mobile user and the M-

commerce application. It is responsible to

communicate with the Mobile Portal,

accept requests from user, and dispatch

corresponding mobile agents to start a trade

transaction. Locating agent provides

discovery services and resides at the

IMAGO discovery server. A vendor agent

acts as the representative of the seller to

keep track of all transactions and inquires.

• Mobile agent: A mobile agent represents

the user roaming the Internet to carry out

the task assigned by its home agent. It will

visit vendors that may carry product desired

by the customer, look for a special service,

and conduct the transaction according to a

specific trading policy.

ICE-B 2009 - International Conference on E-business

3 SERVICE DISCOVERY,

DATABASE SERVICE AND

AGENT COMMUNICATION

A commonly used M-commerce approach is to let

consumers to search for products from physically

distributed vendor sites. However, this approach

may be impractical to a large scale of E-markets

distributed over the Internet. Deploying mobile

agent paradigm in M-commerce offers a possible

solution because the customer may dispatch agents

to search for possible vendor locations. This in turn

leads us to the service discovery problem, that is,

how to find vendor sites available to an M-

commerce application. Following this trend, it

becomes increasingly important to give agents the

ability of finding and making use of services that are

available in a network (Bettstetter and Renner,

2000). A variety of Service Discovery Protocols

(SDPs) are currently under development by some

companies and research groups. The most well

known schemes are Sun's Java based Jini

(Hashman and Knudsen, 2001), Salutation

(Salutation Consortium, 1998), Microsoft's UPP

(Universal Plug and Play Forum, 2006), IETF's

Service Location Protocol (SLP) (Guttman, Perkins,

and Veizades, 1999) and OASIS UDDI (OASIS

UDDI Spec TC, 2005). Some of these SDPs are

extended and applied by several mobile agent

systems to solve the service discovery problem. For

example, GTA/Agent (Rubinstein and Carlos, 1998)

addresses the service location issue by extending

SLP, a simple, lightweight protocol for automatic

maintenance and advertisement of intranet services.

In the IMAGO system, we have implemented a

new service discovery mechanism which is called

Discovery Service via Search Engine Model

(DSSEM) (Song, Li and Ni, 2006). DSSEM is based

on a search engine, a global Web search tool with

centralized index and fuzzy retrieval. Using this

model, E-commerce service providers manually

A mobile agent locates a specific service by

migrating to the service discovery server and

subsequently submitting requests with the required

description. Before a service can be discovered, it

should make itself public. This process is called

service advertisement. A service advertisement

should consist of the service identifier, plus a simple

string describing what the service is, or a set of

strings for specifications and attributes. The most

significant feature of DSSEM is that we enrich the

service description by using web page’s URL to

replace the traditional string-set service description

in mobile agent systems. On the other hand, a

mobile agent can move to a service discovery server,

communicate with the locating agent to obtain an

itinerary that includes a list of ranked host addresses

of the service providers. Based on the given

itinerary, the mobile agent may travel from host to

host to carry out an M-commerce transaction.

In order to let vendor agents communicate with

database systems, the IMAGO system provides a set

of database access primitives, which enables agents

to establish connection with data sources and make

requests for desired information. The system offers

two ways of database accessing, i.e., the set retrieval

and the tuple retrieval. The former returns the entire

matching data set to the requesting agent, whereas

the latter allows the requesting agent to consume the

matching data on the tuple by tuple basis. The

database management module not only provides

flexible interface for accessing data, but also

manipulates database connections efficiently. At the

current stage, the database model in the IMAGO

system is MySQL, the most popular open source

DBMS system in the world.

In an M-commerce application, agents are not

working alone. They need to communicate with each

other to cooperate and generate a global data

aggregation for further analysis. For example,

mobile agents of an M-commerce application might

exchange messages to compare prices of goods they

gathered before making a shopping decision. Most

existing mobile agent systems adopt some kind of

communication models/protocols from traditional

distributed systems. However, the IMAGO system

adopts a different strategy to cope with this issue.

The idea is to deploy intelligent mobile messengers

for inter-agent communication (Li and Autran,

2005). Messengers are thin agents dedicated to

deliver messages. The IMAGO system provides a

set of built-in messengers as a part of its

programming interface. A mobile agent at any

remote sites and at any time may dispatch

messengers to deliver data to designated receivers.

For example, suppose that a mobile agent has

completed its M-commerce work at a remote vendor

server, it can either migrate back to its home server,

or dispatch a messenger to deliver result to the home

agent.

Communication among agents takes place by

means of an Agent Communication Language

(ACL). In order to facilitate open standards of

ACL’s, the IMAGO agent-based communication

model is in compliance with the FIPA ACL message

structure specification (FIPA ACL, 2005). In

addition to the various types of system built-in

messengers for sending agents, the IMAGO system

provides a set of predicates for receiving agents.

USING MOBILE AGENTS TO SUPPORT M-COMMERCE

Typical IMAGO built-in predicates for service

discovery, database access, communication and

agent management are summarized in Table 1.

Table 1: Built-in Predicates.

Category Examples Server type

Service Discovery web_search IMAGO

Discovery server

Database Service db_connection,

db_search_set,

db_search_tuple, etc.

IMAGO

Database server

Communication dispatch, accept, wait_

accept, etc.

IMAGO server

Agent management create, clone, dispose,

move, etc.

IMAGO server

4 SECURE AGENT MIGRATION

CHANNEL

In a large-scale open network, we have to assume

that a mobile agent system is subject to the

fundamental security threats of disclosure,

modification, denial of service, misuse, abuse,

repudiation, and so on. The security challenge here

is much more severe than the one in traditional

client/server model. In addition, security mechanism

should be able fit the purpose of various M-

commerce applications with different security

requirements. Therefore, we are encouraged to

develop the security mechanism based on the most

well known techniques. We may adopt WAP WTLS

(WAP Forum, 2001) or TLS (Allen and Pierks,

1999) to provide message integrity, authentication

and non-repudiation between mobile devices and the

home server. On the other hands, we provide a

simple security mechanism within the execution

environment to protect mobile agents as well as

agent servers. In the current design of the IMAGO

system, the integrated security mechanism consists

of three kernel components: secure migration

protocol, agent verifier and security monitor.

As the primary identifying characteristic of

IMAGO system is the sole protocol to support both

agent mobility and inter-agent communication, we

provide a secure and authenticated connection

between two end points of a migration channel to

protect the privacy and integrity of a moving agent.

The technique incorporated with the IMAGO

migration protocol is the excellent OpenSSL. It is a

full-featured toolkit implementing socket and

transport layer security protocols as well as a set of

general-purpose cryptography algorithms. Based on

OpenSSL, we construct a secure server-to-server

agent migration protocol with ability of

authentication by using the RSA and DSA

algorithms, which are based on a pair of

complementary numerical keys.

When an agent intends to move to another

host, it will be transferred to the Agent Out module.

However, before the agent migration takes place, the

sending server requires some initial configuration if

the receiving server is first met. For each new

destined server, the initial configuration is

performed only once and then cached in a special

internal table. The main purpose of the initialization

is to exchange public keys of both parties and

negotiate the data encryption option. OpenSSL

offers multiple symmetric cryptography algorithms.

For the sake of flexibility and availability, several

cryptography options, such as, from the most secure

to the least secure, 3DES, BlowFish, IDEA, DES

and RC4, will be negotiated with respect to their

preference order, until an option is agreed by both

sending and receiving servers.

The secure migration channel guarantees the

confidentiality, integrity, and password-less

authentication of an agent moving from host to host.

However, the use of mobile agents raises another

two important security concerns, i.e., agents need

protection from malicious hosts and hosts need to be

protected against malicious agents. Unfortunately,

the problem posed by malicious hosts to agents

seems impossible to be fully solved, because there is

no easy way to enforce a host, especially a malicious

host, to obey security requirements. Once an agent

has arrived at a host, it submits itself completely to

the host and cannot stop the host from malicious

attacking. Even though some solutions have been

proposed, many of them are used for attack

detection, rather than prevent agents against

misbehaving servers. At current stage, we assume

that IMAGO M-commerce servers behave truthfully

and honestly.

On the other hand, the IMAGO system takes a

great measure to protect hosts against malicious

agents. First of all, the IMAGO virtual machine is

constructed as a sandbox, a commonly used

technique to control agent execution at the byte-code

instruction level. Secondly, there are two types of

security policies for detecting malicious agents. One

is agent verification and the other is runtime

monitoring. Agent verification is mainly used to

check an incoming mobile agent that has just arrived

from a foreign host. Unfortunately, the verification

process cannot find all potential dangers to the host

server. It is possible that some agents are

deliberately coded to damage remote servers and

some are poorly coded to cause unexpected, harmful

ICE-B 2009 - International Conference on E-business

side effects. Even though the verifier can protect

hosts against illegal instructions, it is not power

enough to prevent denial of service attacks. In order

to prevent such potential risks, a run-time check is

required to monitor the agent’s execution. We adopt

limitation techniques to control the persistent

survivability of mobile agents, that is, to directly

control mobile agents from their dynamic

behaviours during different stages of their lifecycle.

5 AN EXAMPLE OF IMAGO

MOBILE PORTAL

The IMAGO mobile Portal is a program installed on

the handheld device. In fact, the Portal is a

simplified IMAGO IDE, a Java-GUI based program

tailored for mobile devices. Figure 2 gives the

current prototype of the Portal interface. From

which, a mobile customer may communicate with

the home server and invoke a desired M-commerce

application. As an example, suppose a customer

wants to monitor the TSX stock market, he may

invoke the Sniffer application from his Portal. The

application program is given below:

:- home_agent(stock_monitor).

stock_monitor(_) :- wait_accept(Mobile_portal, start),

create(“sniffer.ima”, sniffer,

[[s(“NT”, 26.00), s(“RY”, 43.00)], [s(“SW”,

53.00)]]),

monitor.

monitor :-wait_accept(W, Msg),

notify(W, Msg),

monitor.

notify(_, closed) :- call_back(Mobile_portal, “market

closed”),

disconnect.

notify(W, Msg) :- call_back(Mobile_portal, from(W,

Msg)).

:-end_home_agent(stock_monitor).

:- mobile_agent(sniffer).

sniffer([Buy, Sale]) :- move(imago_discovery_server),

web_search(locate(“TSX”, stock, imago_server), 1,

Server),

move(Server),

split(Buy, Sale).

split([], Sale) :- !, sniff(Sale, sale).

split(Buy, []) :- !, sniff(Buy, buy).

split(Buy, Sale) :- clone(twin, R), R == clone

→

sniff(Sale, sale); sniff(Buy, buy).

sniff(L, Act):- query(L, Act),

sleep(2000),

sniff(L, Act).

Figure 2: The IMA GO Mobile Portal.

query([], _):- !.

query([s(Stock, Limit)|L], Act):- db_search(select(Stock),

_, Value), // database service

check(Stock, Limit, Value, Act),

query(L, Act).

check(_, _, closed, _) :- !, // if market closed

dispatch($oneway_messenger, home, closed),

dispose.

check(Stock, Limit, Value, buy) :- Limit > Value, !,

dispatch($oneway_messenger, home, knock(buy,

Stock, Value)).

check(Stock, Limit, Value, sale) :- Limit < Value, !,

dispatch($oneway_messenger, home, knock(sale,

Stock, Value)).

check(_, _, _, _). // otherwise, take no action

:- end_mobile_agent(sniffer).

The code of stock_monitor defines the home

agent resided at the home stationary server. Upon

receiving the request from the Portal, the

stock_monitor creates a mobile agent called sniffer

alone with initial arguments. When the sniffer starts

execution, it migrates to the imago discovery server

and searches for the URL of TSX IMAGO server

and then moves to that server. Having arrived to the

TSX server, the sniffer continues execution by

calling split/2 which will examine the given

argument list to determine whether a clone is

necessary. If the argument involves both Buy and

Sale stocks, the sniffer clones itself such that the

original sniffs Buy list whereas the clone sniffs the

Sale list. Now, the sniffer will make queries to the

stock vendor agent periodically until the stock

market is closed. For each stock listed in its

argument, the sniffer checks if the new price is less

than the user's limit. If so, an $oneway_messenger is

dispatched to knock the home agent up, otherwise,

the next stock will be investigated. The clone, if

there is one, will do the similar work, except it

checks for the condition on sale. Clearly, it is

possible that no knock-up messengers would be

dispatched ever if the stock prices could not meet the

USING MOBILE AGENTS TO SUPPORT M-COMMERCE

conditions for sale or buy.

6 CONCLUSIONS

In this paper, we discussed the scheme of deploying

mobile agents in M-commerce applications. The

advantage of adopting mobile agents for M-

commerce is to scale up to large, dynamic world

market places distributed over the Internet and to

ease the access and participation of mobile users.

We presented the design of IMAGO M-commerce

framework, discussed the service discovery module

and database management module, and introduced

the work-in-progress IMAGO Mobile Portal and an

excremental example. The API of the IMAGO

system is a set built-in predicates capable to couple a

logic programming language with functionalities of

locating services and accessing remote databases.

Equipped with those system tools, mobile agents

may search for suitable market places, roam the

Internet to collect useful information, compare and

evaluate goods and prices, conduct purchasing

transactions, and communicate with each other to

generate a global view of data through the

aggregation of distributed computations.

Research on the agent based M-commerce

involves further extensions of the IMAGO system.

First, the current design of Mobile Portal should be

further refined to provide an ease-to-use interface

and fully implemented and tested. Secondly, since

E-commerce databases may contain multi-

dimensional data, retrieving such kind of

information from flat web pages is a pending

problem. We are looking to use XML meta-data to

solve the database dimensional problem. In addition,

we are making investigations on adding more

programming languages to the IMAGO system, as

well as introducing more flexible and efficient

communication tools, such as mobile socket, to

facilitate M-commerce applications.

ACKNOWLEDGEMENTS

I would like to express my appreciation to the

Natural Science and Engineering Council of Canada

for supporting this research.

REFERENCES

Shi, N., Ed, 2004. Mobile Commerce Applications,

Hershey PA: Idea Group Publishing

Bădică, C., Ganzha, M., Paprzycki, M., 2005. Mobile

Agents in a Multi-Agent E-Commerce System, In Proc.

Of SYNASC, Timisoara: IEEE Computer Society

Press, pp. 207-214

Bai, L., Chou, D.C., Yen, D.C., Lin, B., 2005. Mobile

Commerce: Its Market Analysis, International Journal

of Mobile Communication, 3(1), 66-81

Kowalczyk, R., Braun, P., Frankczyk, B., Speck, A., 2003.

Deploying Mobile and Intelligent Agents in

Interconnected E-marketplaces, Journal of Integrated

Design and Process Science, 7( 3), 109-123

Li, X., 2006. On the Implementation of IMAGO System,

International Journal of Computer Science and

Network Security, 6( 2a), 107-118

Ngai, E.W.T., Gunasekaran, A., 2007. A Review for

Mobile Commerce Research and Applications,

Decision Support Systems, 43, 3-15

Bettstetter, C., Renner, C., 2000. A Comparison of Service

Discovery Protocols and Implementation of the

Service Location Protocol, In Proc. of EUNICE 2000,

sixth EUNICE Open European Summer School,

Enschede, The Netherlands, pp. 13-15

Hashman, S., Knudsen, S., 2001. The Application of

Jini

Technology to Enhance the Delivery of Mobile

Services,

http://www.sun.com/software/jini/whitepapers/PsiNap

ticMIDs.pdf

Salutation Consortium, 1998. Salutation Architecture

Overview, White Paper,

http:// www.salutation.org/whitepaper/originalwp.pdf

Universal Plug and Play Forum, 2006. UPnP

Device

Architecture, http://www.upnp.org/specs/arch/UPnP-

DeviceArchitecture-v1.0.pdf

Guttman, E., Perkins, C., Veizades, J., 1999. Service

Location Protocol, Version 2, White Paper, IETF,

RFC 2608

OASIS UDDI Spec TC, 2005. Universal Description,

Discovery and Integration v3.0.2 (UDDI),

http://www.oasis-open.org/committees/uddi-

spec/doc/spec/v3/uddi-v3.0.2-20041019.htm

Rubinstein, M., Carlos, O., 1998. Service Location for

Mobile Agent System, In IEEE/SBT International

Telecommunications Symposium (ITS'98), SP Brazil,

pp. 623-626.

Song, L., Li, X, Ni, J., 2006. A Database Service

Discovery Model for Mobile Agents, International

Journal of Intelligent Information Technologies, 2(2),

16-29

Li, X., Autran, G., 2005. Inter-agent Communication in

IMAGO Prolog, Lecture Notes in Artificial

Intelligence,

3346, 163-180

FIPA ACL, 2005. Agent Communication Language

Specifications, FIPA,

WAP Forum, 2001. WAP WTLS: Wireless Application

Protocol Wireless Transport Layer Security

Specification,

http://www1.wapforum.org/tech/terms.asp?doc=OMA

-WAP-260_101-WIM-SIN-20020107-a.pdf

Allen, C., Pierks, T., 1999. The TLS Protocol Version 1.0,

RFC2246, http://www.ietf.org/rfc/rfc2246.txt

ICE-B 2009 - International Conference on E-business