844 Applications of Agent-Based Technology as Coordination and Cooperation • How much does the agent need to reduce the price to win the bid? To improve the performance of the agent, it is necessary to learn from the history of the game. For example, Figure 4 presents the average price of PC of the competition. The agents can learn from the chart when the market price of PCs are high, medium, and low. Equilibrium prices arise when supply equals demand: Q s i = Q d i for product i. If Q s i Q d i , agents will bid price P i lower ; if Q s i Q d i , agents will bid price P i higher. Usually the price of the product increases at the begin- ning of game due to lack of supplies. Therefore, the agents who supply the product at the time of low market supply can get a higher price and, as a result, can earn more market share with more S UR¿W & RQ VHT XH QW O\ WKH D JH QW Z KR F DQ D GR S WW K L V strategy of increased productivity, and bids ac- cording to the market situation will have a better RSSRUWXQLW\WRPD[LPL]HSUR¿W Huq (2006) analyzed the product market of the TAC/SCM 2004 game and observes the lack of cooperation among agents involved in component purchasing and product selling. The DYHUDJHPDUNHWGHPDQGIRU3&VLQWKHVHPL¿QDO DQG¿QDOURXQGJDPHFDQEHGHSLFWHGLQ7DEOHV 1 and 2, where the second column is the average 7DEOH$YHUDJHWRWDO3&VGHOLYHUHGLQVHPL¿QDORI*U7$&DQG7$& Agents Average Delivery Total Average Price % of Order Average FreeAgent 46882 291505 16 1656 SouthamptionSCM 61759 21 1508 Mr. UMBC 51587 18 1527 ScrAgent 40995 14 1504 KrokodilAgent 41551 14 1538 Socrates 48732 17 1323 7DEOH$YHUDJHWRWDO3&VRUGHUHGE\DJHQWVLQ¿QDOURXQG Agents Average PC Delivery Total Average Market Demand % of Order Price Average FreeAgent 41659 201227 21 1842 SouthamptionSCM 45465 23 1670 Mr. UMBC 44665 22 1481 ScrAgent 13765 7 1434 KrokodilAgent 24487 12 1869 Socrates 31186 15 1764 845 Applications of Agent-Based Technology as Coordination and Cooperation PCs delivered by the agents; the third column is the total average market demand. The authors VXEVHTXHQWO\¿QGWKDWWKHIUHHDJHQWELGVRQDQ average with a higher average price and a higher percentage of orders. In summary, the TAC SCM has a distinct lack of cooperation among the agents involved in component purchasing and product selling, and this led the authors to conclude that the game was a likely case study to investigate modeling coordination and cooperation. MODELING COORDINATION AND COOPERATION IN TAC/SCM According to the TAC/SCM, all manufacturer agents are rational or self-interested and their PDLQIRFXVLVWRPD[LPL]HSUR¿W,IZHDVVXPH that the agents cannot achieve their goal in isola- tion or that they would prefer to work with each other, then this has the potential for cooperation. In this context, all the manufacturer agents can work together towards their goal. On the one hand, manufact urer agents will be able to increase their SURGXFWLRQFDSDFLW\DQGVHOOWKH¿QDOSURGXFWVWR customers, and on the other hand, suppliers will EHQH¿W E\ VXSSO\LQJ PRUH FRPSRQHQWV WR WKH PDQXIDFWXUHUVZKLFKZLOOUHVXOWLQPRUHSUR¿W The following discussion proposes a theoretical model, which will be able to solve the coordina- tion and cooperation problem of the TAC/SCM game. The authors have found in the TAC/SCM com- petition that three or four agents always dominant the market of buying components or selling the products. Therefore, this research characterized these agents as big agents and the other agents as small/medium agents (SMAs). Again, it was also found that SMAs could not purchase enough V K D UH R I W K HF R PS RQ H Q WV WR SU R G XF HD ¿ QD OS U R GX FW WRVHOO7KLVLVDWHFKQLFDOVWUDWHJLFRU¿QDQFLDO problem for the SMAs. Consequently, if the SMAs purchase components from big agents and sell to customers, then it is possible to survive. Otherwise, the SMAs cannot compete with the big agents. In the real world, usually the intention of large organizations is to extend their business DQGPDNHPRUHSUR¿W7KLVLQFUHDVHVSURGXFWLRQ ZKLFKXOWLPDWHO\OHDGVWRLQFUHDVHGSUR¿W8VLQJ this strategy, we assume that big agents want to extend their business, and at the same time, the SMAs would like to work with big agents. This way, big agents and SMAs can work together to achieve their common goals. As a result, every DJHQWZLOOEHEHQH¿WHGE\SDUWLFLSDWLQJLQVKDUHG activities. Therefore, to work together the agents QHHGWRIROORZWKHVWDJHVGH¿QHGLQWKHSUHYLRXV section, Cooperative Problem Solving Process. In this regard, the following characteristics can EHGH¿QHGVHHWKHDJHQWW\SHVLQ'H¿QLWLRQRI the Recognition Stage of the Cooperative Problem Solving Process section): Theorem 6. a. There exist some group of agents g such that the individual agent i believes that the g can jointly achieve goal. b. either: c. An agent i cannot achieve goal individu- ally. d. an agent i believes for every action that could be performed to achieve the task, it has a goal of not performing the goal. Theorem 7. 7KHRXWFRPHVHQVXUHWKHLUSUR¿WLI and only if the cooperative agents complete their task successfully. If the cooperative agents complete their task successfully, then all the participating agents will VKDUHWKHSUR¿WRWKHUZLVHLWZLOOEHFRQVLGHUHG an incomplete task. Theorem 8. The cooperative agents are those if and only if they agree to work together. 846 Applications of Agent-Based Technology as Coordination and Cooperation In the Cooperative Processing Stage, only those able agents that are determined to complete their tasks towards a common goal are considered cooperative agents. 'H¿QLWLRQA decommitted agent is an agent that started its task but did not complete that task, and therefore needs to be penalized. 'H¿QLWLRQ Let a set of able agents that share their work to achieve a common goal be called cooperative agents, which is: ag i Ȏ A ={ ag 1 , ag 2 , …….ag n }= I (1) 'H¿QLWLRQ The accumulated task of the coop- erative agents A, the utility u of that task can be considered as unique, and can be expressed as: u(A) = 6 A i = n1 (2) i = 1 'H¿QLWLRQ3UR¿WDOORFDWLRQWRWKHDJHQWV The percentage of the utility of each agent can be worked out according to the contribution of each agent, which can be expressed as: ( ) 100 () () i i uag uag uA u (3) 'H¿QLWLRQ Cooperative action takeover: If any agent fails to complete its task, then other agents will need to complete that task to achieve the goal. ,IDQ\DJHQWLVXQDEOHWR¿QLVKLWVDOORFDWHG tasks due to unavoidable circumstances, then the RWKHUDJHQWVZLOOWDNHRYHUWKDWXQ¿QLVKHGWDVN enthusiastically to achieve the goal. 'H¿QLWLRQThe set of cooperative agents A are D¿QLWHVHWDQGVDLGWREHERXQGHG The cooperative agents must be limited in QXPEHUIRUHI¿FLHQF\LQWDVNDOORFDWLRQDVLWLV n o t p o s s i bl e t o h a ve a n u n l i m i t e d n u m b e r o f a g e n t s working together. The cooperative agents are bounded, for instance: (a) the agent who invests or sells the greatest is called the upper bounded; and (b) the agent who invests or sells the lowest is called the lower bounded. Architecture of the Cooperative Processing Agents Let us consider that a number of companies in different locations have agreed to sell some prod- ucts to customers within a limited time frame. Assume that the agents are going to work together according to the Cooperative Processing Stages. A proposition for architecture of effective coop- erative processing is shown in Figure 5. In this ¿J X U H W KH UH L V D FRO OH F W LR Q RI PD QX ID F W X UH UD JH QW V n in the domain. When these agents have agreed WR SHUIRUP WDVNV WR DFKLHYHD VSHFL¿F JRDO WR complete the cooperative processing, other agents are needed. This research argues that these agents are Task Allocation Agent, Monitoring Agent, Evaluating Agent, Result Allocation Agent, and Coordination Manager Agent (CoManager). When a problem is decomposed into smaller subproblems, the Task Allocation Agent is re- sponsible for allocating tasks to the able agents in order to achieve the goal. The Monitoring Agent is responsible for monitoring the performance of the agents’ tasks, that is, which agent is doing its task and which is not. Finally, this agent will produce a report to the CoManager Agent. According to this report, the CoManager Agent will reallocate WKHXQ¿QLVKHGWDVNWRWKHDJHQWWKDWLVZLOOLQJWR undertake that task. The Evaluating Agent will evaluate all tasks from the Monitoring Agent. The Evaluating Agent will provide analytical and objective feedback on HI¿FLHQF\DQGHIIHFWLYHQHVVRIWKHSHUIRUPDQFH 847 Applications of Agent-Based Technology as Coordination and Cooperation RIDJHQWV)LQDOO\LWZLOOSURGXFHDQRYHUDOO¿QDO UHSRUWLQFOXGLQJEHQH¿WVRIHDFKDJHQWWRWKH Result Allocation Agent (YHQWXDOO\ WKLV ¿QDO report allows the agents to learn lessons. The Result Allocation AgentWKHQSURFHVVHVWKHEHQH¿WV GHVHUYHGE\HDFKDJHQWDQG¿QDOO\SURGXFHVD EHQH¿WUHSRUWWRWKHDJHQWV The contribution made by this research is the addition of the monitoring and evaluation stages for the Cooperative Problem Solving Process, and the results described in this section. The TAC/SCM was used as a case study to illustrate WKHFRQFHSWVRXWOLQHGLQWKHWKHRUHPVDQGGH¿QL- tions. FUTURE TRENDS The potential for B2B e-commerce is now pro- jected to be much larger than that for consumer oriented e-commerce (Chan, Lee, Dillon, & Chang, 2001). Conducting electronic B2B trans- actions is an emerging and potentially lucrative issue. For example, in the supply chain, manu- facturer organizations or retailers are dependent on supplier organizations. There are many processes in selling and purchasing that can be conducted electronically. Particularly, at the time of purchasing, many processes are complex and involve negotiation, cooperation, and coordina- tion. In the real world, these processes are very time consuming and complicated. Therefore, if Figure 5. Architecture of effective cooperation model 848 Applications of Agent-Based Technology as Coordination and Cooperation we can utilize these processes electronically, we can avoid complexity and will be able to reduce costs and time taken. Figure 6 shows how B2B e-commerce has grown from 1998 to 2005. As a result, we can predict that this trend in e-business utilization will increase into the future. As described in the previous sections, both large organizations and SMEs will be able to work to- gether to conduct e-business on a global basis. In regards to implementing cooperative work utiliz- ing multi-agent systems, the agents need to follow WKHVWDJHVGH¿QHGLQWKLVFKDSWHU$VDUHVXOWDOO WKHSDUWLFLSDQWRUJDQL]DWLRQVZLOOEHQH¿WLQRYHUDOO performance outcomes. In conclusion, the authors argue that team effort, rather than individual ef- fort, will give more robust and sustainable results. The cooperation and coordination protocol, and information sharing among various agents can be future research areas, which will facilitate in building the software that enables coordination and cooperation activities. CONCLUSION 7KLVFKDSWHULGHQWL¿HGSUREOHPVLQFRQGXFWLQJ e-business and managing the supply chain. It DOVRLGHQWL¿HGH[SHFWHGEHQH¿WVIRUVXSSO\FKDLQV with agents working together in coordinated and cooperative processes. The utilization of a multi-agent system in supply chain management and the cooperative problem solving stages have been presented and discussed. To apply these stages, the proposition for architecture of effec- tive cooperative processing for agents and some characteristics in modeling coordination and Figure 6. Projection growth of B2B e-commerce drawn from a report by Gartner Group Growth of B2B E-Commerce 45 145 403 953 2188 3949 7297 8500 0 1500 3000 4500 6000 7500 9000 1998 1999 2000 2001 2002 2003 2004 2005 Year Billions of Dollars 849 Applications of Agent-Based Technology as Coordination and Cooperation cooperation for TAC/SCM have been outlined. The ultimate goal is to develop the capability of organizations to work effectively together in online e-business transactions. In addition to this, large organizations can expand their businesses and SMEs can work with large organizations. Finally, it can reduce time for selling and buying DFWLYLWLHVDQGLQFUHDVHWKHWRWDOSUR¿WVRIWKHVXS- ply chain. In addition, it will facilitate the ability to cooperate and coordinate among multi-agents in e-commerce. Further, it will enhance customer satisfaction and streamline B2B transactions by reducing transaction costs of tasks at every stage of the supply chain. Therefore, it will increase WUXVW DQG FRQ¿GHQFH LQ WKH FRPSRQHQW PDUNHW and product market. REFERENCES Beck, J. C., & Fox, M. S. (1994, May 15). Sup- ply chain coordination via mediated constraint relaxation. Paper presented at the First Canadian :RUNVKRSRQ'LVWULEXWHG$UWL¿FLDO,QWHOOLJHQFH Banff. Becker, R., Holland, O. E., & Deneubourg, J. L. (1994). From local actions to global tasks: Stig- mergy in collective robotics. In R. Brooks & P. Maes (Eds.), $UWL¿FLDO/LIH,9. Cambridge, MA: MIT Press. Chan, H., Lee, R., Dillon, T., & Chang, E. (2001). E-commerce fundamentals and applications. West Sussex, England: John Wiley & Sons. Chopra, S., & Meindl, P. (2003). Supply chain management: Strategy, planning, and operation (2 nd ed.). Collins, J., Arunachalam, R., Sadeh, N., Eriksson, J., Finne, N., & Janson, S. (2005). The supply chain game for the 2006 trading agent competition, competitive benchmarking for the trading agent community. Retrieved August 18, 2007, from http://www.sics.se/tac/tac06scmspec_v16.pdf Doran, J. E., & Palmer, M. (1995). The EOS Project: Integrating two models of palaeolithic social change. In N. Gilvert & R. Conte (Eds.), $UWL¿FLDO6RFLHWLHV7KH&RPSXWHU6LPXODWLRQRI Social Life (pp. 103-105). London: UCL Press. Finnie, G., Berker, J., & Sun, Z. (2004, August). A multi-agent model for cooperation and negotia- tion in supply networks. Paper presented at the Americas Conference on Information Systems, New York. Finnie, G., & Sun, Z. (2003). A knowledge-based model of multiagent CBR systems. Paper presented at the International Conference on Intelligent Agents, Web Technologies, and Internet Com- merce (IAWTIC’2003), Vienna, Austria. Fox, M. S. (1981). An organizational view of dis- tributed systems. IEEE Transactions on Systems, Man and Cyvernetics, 11(1), 70-79. Franklin, S. (1996). Coordination without com- munication. Retrieved August 18, 2007 from http://www.msci.memphis.edu/~franklin/coord. html Franklin, S., & Graesser, A. (1997). Is it an agent, or just a program? A taxonomy for autonomous agents. Paper presented at the Third International Workshop on Agent Theories, Architectures, and Languages. +HZLWW&2I¿FHVDUHRSHQV\VWHPVACM 7UDQVDFWLRQVRQ2I¿FH6\VWHPV(3), 271-287. Holt, A. W. (1988). Diplans: A new language for the study and implementation of coordination. $&0 7UDQVIRUPDWLRQV RQ 2I¿FH ,QIRUPDWLRQ Systems, 6(2), 109-125. Huberman, B. A. (1988 ). The ecology of compu- tation. Amsterdam: North-Holland. Huq, G. B. (2006, February 13-20). Analysis, planning and practice of trading agent competi- tion supply chain management (TAC/SCM). Paper presented at the 2 nd International Conference on Information Management Business, Sydney. 850 Applications of Agent-Based Technology as Coordination and Cooperation Jennings, N. R. (1990). Coordination techniques IRUGLVWULEXWHG DUWL¿FLDOLQWHOOLJHQFH,Q*03 O’Hare & N.R. Jennings (Ed.), Foundations of 'LVWULEXWHG$UWL¿FLDO,QWHOOLJHQFH (pp. 187-210). London: Wiley. Jennings, N. R. (2000) On agent-based software engineering. $UWL¿FDO ,QWHOOLJHQFH (2), 277- 296. Melone, T. W., & Crowston, K. (1990, October 7-10). What is coordination theory and how can it help design cooperative work systems? Paper presented at the ACM Conference on Computer Supported Cooperative Work (CSCW), Los Angeles. Meyer, R. A. (1976). Microeconomic decisions. +RXJKWRQ0LIÀLQ&RPSDQ\ Miller, M. S., & Drexler, K. E. (1988). Markets and computation: Agoric open systems. Amsterdam: North-Holland. In B.A. Huberman (Eds.), The Ecology of Computation (pp. 133-176). Nwana, H. S. (1994). Negotiation strategies: An overview (BT Laboratories internal report). Nwana, H. S., Lee, L., & Jennings, N. (1996). Co-ordination in software agent systems. British Telecom Technical Journal, 14(4), 79-88. Rosenschein, J. S., & Zlotkin, G. (1994). Rules of encounter: Designing conventions for automated negotiation among computers. Cambridge: MIT Press. Schneider, P. G., & Perry, J. T. (2001). Electronic Commerce. In Course Technology. Canada. Sichman, S. J. (1994). A social reasoning mecha- nism based on dependence networks. In Proceed- ings of the 11 th (XURSHDQ&RQIHUHQFHRQ$UWL¿FLDO Intelligence (ECAI-94), Amsterdam. Sichman, S. J., & Demazeau, Y. (1995). Exploiting social reasoning to deal with agency level incon- sistency. Paper presented at the 1 st International Conference on Multi-Agent Systems (ICMAS-95), San Francisco. Smith, R. G., & Davis, R. (1981). Frameworks for cooperation in distributed problem solving. IEEE Transactions on Systems, Man and Cyvernetics, 11(1), 61-70. Sycara, K. P. (1989). Multiagent compromise via negotiation. In L. Gasser & M. Huhns (Eds.), 'LVWULEXWHG$UWL¿FLDO,QWHOOLJHQFH(Vol. II, , pp. 119-138). London: Morgan Kaufmann/San Mateo, CA: Pitman Publishing. Winogard, T., & Flores, F. (1986). Understanding computers and cognition: A new foundation for design. Noorwood, NJ: Ablex. Wooldridge, M. (2002). An introduction to mul- tiagent systems.: John Wiley & Sons. Wooldridge, M., & Jennings, N. R. (1995). Intel- ligent agents: Theory and practice. Knowledge Engineering Review, 2(10), 115-152. Wooldridge, M., & Jennings, N. R. (1999). The cooperative problem-solving process. Journal of Logic Computation, 9(4), 563-592. This work was previously published in Agent Systems in Electronic Business, edited by E. Li and S. Yuan, pp. 125-145, copyright 2008 by Information Science Reference (an imprint of IGI Global). 851 Copyright © 2009, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited. Chapter 3.12 Secure Agent Roaming for Mobile Business Sheng-Uei Guan National University of Singapore, Singapore ABSTRACT M-commerce, a new way to conduct business, is gaining more and more popularity due to the wide use of the Internet. Despite its rapid growth, there are limitations that hinder the expansion of m-commerce. The primary concern for online shopping is security. Due to the open nature of the ,QWHUQHWSHUVRQDO¿QDQFLDOGHWDLOVQHFHVVDU\IRU R Q O L Q HV K RS SL Q J FD QE HV W RO HQ L IV XI ¿F LH QW VH FX U LW \ mechanism is not put in place. How to provide the necessary assurance of security to consumers remains a question mark despite various past ef- forts. Another concern is the lack of intelligence in locating the correct piece of information. The Internet is an ocean of information depository. It is rich in content but lacks the necessary intel- ligent tools to help one locate the correct piece of information. Intelligent agent, a piece of software that can act intelligently on behalf of its owner, is GHVLJQHGWR¿OOWKLVJDS+RZHYHUQRPDWWHUKRZ intelligent an agent is, its functionality is limited if it remains on its owner’s machine and does not have any roaming capability. With the roaming ca- p a b i l i t y, m o r e s e c u r i t y c o n c e r n s a r i s e . I n r e s p o n s e to these concerns, SAFE, Secure roaming Agent For E-commerce, is designed to provide secure roaming capability to intelligent agents. INTRODUCTION The introduction of the Internet is probably one RIWKHPRVWVLJQL¿FDQWUHYROXWLRQVRIWKH th century. With a simple click, one can connect to almost every corner of the world thousands of kilometers away. This presents a great opportunity for mobile commerce (m-commerce). Despite its many advantages over traditional commerce, m- commerce has not taken off successfully. One of the main hindrances is security. When it comes to online transactions, security becomes the primary concern. The Internet was developed without too much security in mind. ,QIRUPDWLRQ ÀRZV IURP KXEV WR KXEV EHIRUH LW reaches the destination. By simply tapping into ZLUHVRUKXEVRQHFDQHDVLO\PRQLWRUDOOWUDI¿F transmitted. For example, when Alice uses her 852 Secure Agent Roaming for Mobile Business VISA credit card to purchase an album from Virtual CD Mall, the information about her card may be stolen if it is not carefully protected. This information may be used maliciously to make other online transactions, thus causing damage to both the card holder and the credit card company. Besides concerns on security, current m-com- merce lacks intelligence. The Internet is like the world’s most complete library collections unsorted by any means. To make things worse, there is no competent librarian that can help readers locate the book wanted. Existing popular search engines are attempts to provide librarian assistance. However, as the collection of information is huge, none of the librarians are competent enough at the moment. The intelligent agent is one solution to pro- viding intelligence in m-commerce. But having DQDJHQWWKDWLVLQWHOOLJHQWLVLQVXI¿FLHQW7KHUH are certain tasks that are unrealistic for agents to perform locally, especially those that require a large amount of information. Therefore, it is important to equip intelligent agents with roam- ing capability. Unfortunately, with the introduction of roam- ing capability, more security issues arise. As the agent needs to move among external hosts to perform its tasks, the agent itself becomes a target of attack. The data collected by agents may EHPRGL¿HGWKHFUHGLWFDUULHGE\DJHQWVPD\EH stolen, and the mission statement on the agent may be changed. As a result, transport security is an immediate concern to agent roaming. SAFE transport protocol is designed to provide a se- cure roaming mechanism for intelligent agents. Here, both general and roaming-related security concerns are addressed carefully. Furthermore, several protocols are designed to address different requirements. An m-commerce application can choose the protocol that is most suitable based on its need. Background on Agents There has been a lot of research in the area of intelligent agents. Some literature only proposes certain features of intelligent agents, some at- WHPSWVWRGH¿QHDFRPSOHWHDJHQWDUFKLWHFWXUH Unfortunately, there is no standardization in the various proposals, resulting in vastly different agent systems. Efforts are made to standardize some aspect of agent systems so that different sys- tems can inter-operate with each other. In the area of knowledge representation and exchange, one of the most widely accepted standards is KQML (Knowledge Query and Manipulation Language) (Finin & Weber, 1993), developed as part of the Knowledge Sharing Effort. KQML is designed as a high-level language for runtime exchange of information between heterogeneous systems. Un- f o r t u n a t e l y, KQ M L i s d e s i g n e d w i t h l i t t l e s e c u r i t y considerations because no security mechanism is built to address common security concerns, not WRPHQWLRQVSHFL¿FVHFXULW\FRQFHUQVLQWURGXFHG by mobile agents. Agent systems using KQML will have to implement security mechanisms on top of KQML to protect them. In an attempt to equip KQML with ‘built-in’ security mechanisms, Secret Agent is proposed by Thirunavukkarasu, )LQLQDQG0D\¿HOG 6HFUHW$JHQWGH¿QHVDVHFXULW\OD\HURQWRS of KQML. Applications will have to implement some special message format in order to make use of Secret Agent. Secret Agent has a number of shortcomings and is handicapped by the design of KQML. Firstly, one requirement of Secret Agent is that every agent implementing the security algorithm must possess a key (master key). This master key is either a symmetric key or based on PKI. If the key is based on a symmetric key algorithm, it requires each agent to have a separate key with every other agent it corresponds with. If the agent intends to communicate with another agent that it has no common pre-established master 853 Secure Agent Roaming for Mobile Business key, a central authentication server is required to generate such a key. The problems introduced are key database management, authentication server protection, and key transport/exchange security. If the master key is based on PKI, the agent identity must be tightly tied with the key pair. 7 K LVZD VL Q VXI ¿FLH QW O\D GG UH VVH G L Q W KHGHVLJ QRI Secret Agent, subjecting the algorithm to man-in- the-middle attack. In the SAFE transport protocol, agent identity and key pair are tightly integrated XVLQJGLJLWDOFHUWL¿FDWLRQ Another prominent transportable agent system is Agent TCL, developed at Dartmouth College (Gray, 1997; Kotz, Marzullo, & Lauvset, 1997). Agent TCL addresses most areas of agent transport by providing a complete suite of solutions. It is probably one of the most complete agent systems under research. Its security mechanism aims at protecting resources and the agent itself. Since some existing agent systems are already very VWURQJLQWKLVDUHD$JHQW7&/³VHHNVWRFRQ¿UP WKHLUVXI¿FLHQF\DQGHLWKHUFRS\RUUHGHVLJQDV appropriate” (Gray, 1997). In terms of agent pro- tection, the author acknowledges that: …it is clear that it is impossible to protect an agent from the machine on which the agent is executing…it is equally clear that it is impossible to protect an agent from a resource that willfully provides false information. (Gray, 1997) As a result, the author: «VHHNVWRLPSOHPHQWDYHUL¿FDWLRQPHFKDQLVPVR that each machine can check whether an agent ZDVPRGL¿HGXQH[SHFWHGO\DIWHULWOHIWWKHKRPH machine. (Gray, 1997) In other words, it addresses agent integrity and provides a certain level of traceability to the agents. The other areas of security—like non- UHSXGLDWLRQYHUL¿FDWLRQDQGLGHQWL¿FDWLRQ²DUH not carefully addressed. Compared with the various agent systems discussed above, SAFE is designed to address the special needs of m-commerce. The other mobile DJHQWV\VWHPVDUHHLWKHUWRRJHQHUDORUWRRVSHFL¿F to a particular application. By designing SAFE with m-commerce application concerns in mind, the architecture will be suitable for m-commerce applications. The most important concern is se- curity, as already discussed. Due to the nature of m-commerce, security becomes a prerequisite for any successful m-commerce application. Other FRQFHUQVDUHPRELOLW\HI¿FLHQF\DQGLQWHURSHU- DELOLW\,QDGGLWLRQWKHGHVLJQDOORZVFHUWDLQÀH[- ibility to cater to different application needs. GENERAL AGENT TRANSPORT As a prerequisite, each SAFE entity must carry DGLJLWDOFHUWL¿FDWHLVVXHGE\6$)(&HUWL¿FDWH Authority, or SCA. In this way, each agent, agent owner, and host will carry its own unique digital FHUWL¿FDWH7KHFHUWL¿FDWHLWVHOILVXVHGWRHVWDEOLVK the identity of a SAFE entity. Because the private NH\WRWKHFHUWL¿FDWHKDVVLJQLQJFDSDELOLW\WKLV DOORZVWKHFHUWL¿FDWHRZQHUWRDXWKHQWLFDWHLWVHOIWR the SAFE community. An assumption is made that the agent private key can be protected by function hiding (Thomas, 1998; other techniques are also discussed in Bem (2000) and Westhoff (2000), but will not be elaborated in this chapter). From the host’s viewpoint, an agent is a piece of foreign code that executes locally. In order to prevent a malicious agent from abusing the host resources, the host should monitor the agent’s usage of resources (e.g., computing resources, network resources). Agent receptionist will act as the middleman to facilitate and monitor agent communication with external party. . processes in selling and purchasing that can be conducted electronically. Particularly, at the time of purchasing, many processes are complex and involve negotiation, cooperation, and coordina- tion E-commerce fundamentals and applications. West Sussex, England: John Wiley & Sons. Chopra, S., & Meindl, P. (2003). Supply chain management: Strategy, planning, and operation (2 nd ed.). Collins,. modeling coordination and cooperation. MODELING COORDINATION AND COOPERATION IN TAC/SCM According to the TAC/SCM, all manufacturer agents are rational or self-interested and their PDLQIRFXVLVWRPD[LPL]HSUR¿W,IZHDVVXPH that