A Multi-Agent System for Content Trading in Electronic Telecom Markets Using Multi-Attribute Auctions Ana Petric Faculty of Electrical Engineering and Computing, University of Zagreb Unska 3, Zagreb, Croatia ana.petric@fer.hr ABSTRACT Generation Network (NGN) provide connections which enable a The advent of the Internet and the development of the New particular lifestyle that is aspiring to digital humanism where Generation Network (NGN) has enabled, while investments in people’s daily activities are becoming more digitalized, licenses and the desire to stay competitive in the future has convenient and intelligent [27]. Actors on the telecom markets are triggered the development of value added services (VAS). Due to pursuing innovations and launching new value-added services high market penetration, the telecommunication industry has been (VAS) [5] in order to increase revenue. This is due to the fact that facing income stagnation. Consequently, it has been shifting focus provisioning basic telecommunication services (i.e., fixed and to VAS in order to increase income. When forming VAS, special mobile communication, data transfer) is no longer enough to keep attention needs to be paid to the purchase of resources (e.g., existing customers, let alone attract new ones, due to high market transport capacity and information resources) needed for the penetration. Investment regain of licenses and staying competitive service creation. The fact that information resources (i.e., content) in the future are key drivers for the expansion of new VAS on the are not commodities, opens the question of what is the best (i.e., market. This new market demand and technological development efficient) mechanism that should be used for trading. As the has led to the convergence of different domains (i.e., number of participants on the B2B telecom market increases, the telecommunications, information technology (IT), the Internet, need for the automation of transactions carried between them is broadcasting and media) all involved in the telecom service critical. The automation of transactions should lower operational provisioning process. The ability to transfer information costs and speed up the service provisioning process. In this paper, embodied in different media into digital form to be deployed we try to identify stakeholders on the telecom e-market, establish across multiple technologies is considered to be the most their roles and relationships and find an appropriate model which fundamental enabler of convergence [14]. An important feature of captures their transactions. Finally we consider the use of multi- convergence is the composition of services and content derived by attribute auctions for content trading in telecom markets. combining multiple simpler services or types of content in order to provide more powerful services. Categories and Subject Descriptors The research problem addressed in this paper concerns the I.2.11 [Distributed Artificial Intelligence]: Intelligent agents, automation of business processes related to the creation of VAS Multiagent systems. J.4. [Social and Behavioral Sciences]: that are traded on the telecommunication electronic markets (e- Economics. I.6.5 [Model Development]: Modeling markets). There are two types of resources needed for the creation methodologies of telecom VAS. They are the information resources (i.e., content) the service is based on and the transport capacities needed for service provisioning. The telecom market is divided into two General Terms submarkets, the B2B (Business-to-Business) market and the B2C Management, Design, Economics. (Business-to-Consumer) market. Our research is focused on B2B telecom e-market trading with information resources using multi- Keywords attribute auctions. Multi-attribute auctions, Multimedia Content Trading, B2B e- The rest of the paper is structured as follows. Section 2 describes markets, Intelligent Software agents, New Generation Network the participants on the telecom e-market. Section 3 describes the phases we need to go through in order to conduct a transaction on 1. INTRODUCTION the B2B telecom e-market. Section 4 addresses general auction The advent of the Internet and the development of the New mechanisms and presents multi-attribute auctions. Section 5 states the main questions of this research effort and proposes some answers. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that 2. TELECOM E-MARKETS copies bear this notice and the full citation on the first page. To copy The appearance of new stakeholders on the B2B telecom market otherwise, or republish, to post on servers or to redistribute to lists, had to be taken into account so new business models were requires prior specific permission and/or a fee. formed. One of the long-term objectives of the NGN is to support 10th Int. Conf. on Electronic Commerce (ICEC) ’08 Innsbruck, Austria business models that open the market to emerging service Copyright 2008 ACM 978-1-60558-075-3/08/08 ...$5.00. providers [14]. The volume and dynamic nature of VAS offered in the NGN place novel demands and challenges on telecom 3. B2B TELECOM E-MARKET stakeholders. In this newly developed situation it is not enough The BBT (Business-to-Business Transaction) model [15] just to adequately respond on the existing requests but also to systematically analyses processes in B2B e-markets. The intelligently anticipate the development of the future events and proliferation of auctions on the Internet, and the dynamic nature adapt to their environment. In order to understand the of auction interactions, argues for the development of intelligent relationships between stakeholders and the way they interact it is trading agents which act on behalf of human traders (i.e., buyers important that their roles are well classified. We use the and sellers). Intelligent trading agents can also be used to classification determined in [8] as shown in Figure 1. impersonate stakeholders in the environment of the NGN in order Consumers are service users that have at their disposal various to enable automated interactions and business transactions on the devices (e.g., mobile phone, laptop, PDA) and are connected telecom markets [20]. Namely, an agent can monitor and through various access networks (e.g., 3G, WiMax). Access participate in the market continuously. Software agents [7, 19] are Provider ensures telecommunication access for service programs which autonomously act on behalf of their principal consumers. Service Provider facilitates a variety of basic and while carrying out complex information and communication tasks integrated services for consumers enabling easy content that have been delegated to them. A software agent is intelligent consumption. Carriers provide a transport service for the data (its intelligence is grounded on its knowledge base, reasoning traffic and they usually buy bandwidth from Network mechanisms and learning capabilities), autonomous, reactive, Infrastructure Owners who provide transmission lines. A large proactive, cooperative, and persistent. Additionally, a software number of Carriers are at the same time also Network agent can also be mobile. Infrastructure Owners. Wholesaler of Capacity provides lower- From the BBT model perspective [15], we can formally identify cost transmission and storage capacity. Content Owner possesses six fundamental steps which must be executed in order to the information in its original format while Content Enabler successfully complete one transaction in a B2B environment. converts this information to a format eligible for the transmission These steps are as follows (Figure 2): 1) partnership formation, 2) over heterogeneous networks. Content Provider is at the same brokering, 3) negotiation, 4) contract formation, 5) contract time Content Owner and Content Enabler. Wholesaler of Content fulfillment, and 6) service and evaluation. B2B negotiation is provides lower-cost content. Server Infrastructure Owner complex since it typically involves larger volumes, repeated provides storage capacity and server functionality. Information transactions and more complicated contracts. This is the reason Enablers enable information resources while Transport Enablers why most researchers have concentrated on the negotiating phase provide transport of information resources through the various of B2B market transactions. networks swiftly and seamlessly. The partnership formation phase usually includes forming of a We are focusing on the B2B e-market since it is widely believed new virtual enterprise or finding partners to form a supply chain. that it will become the primer way of doing business [21]. The A virtual enterprise represents a form of cooperation of assumption is that the telecom B2B e-market will grow with other independent stakeholders which combine their competencies in B2B e-markets. A special intention is paid to the negotiation order to provide a service [6]. On the B2B telecom e-market, phase since the outcome (i.e. financial efficiency) is still the Content Owners, Content Enablers, Server Infrastructure Owners premier performance measure for most businesses [16, 17]. and Wholesalers of Content can form a virtual enterprise in order to successfully place and sell information resources to various service providers. Moreover, Carriers, Network Infrastructure USE Owners and Wholesalers of Capacity may also form a Virtual n n Enterprise to enhance trading with transport capacity. With the n expansion of the e-market, the number of buyers and sellers FACILITATE grows accordingly making it more difficult to find all potential n business partners trading a requested service/resource. The main role of the Brokering phase is to match service providers with information/transport enablers that sell information resources/transport capacities needed for the creation of a new n n service or improvements of an old one. STORE & ENABLE TRANSPORT Negotiation is a process which tries to reach an agreement n n regarding one or more resource attributes (e.g., price, quality, etc.). Each stakeholder in the negotiation process is represented by an intelligent trading agent that negotiates in his behalf (e.g., Information Agent trades in behalf of Information Enabler). The Content Owner n Carrier n trading agent uses a negotiation strategy suitable for the type of Content Enabler n Network Infrastrucutre auction applied (i.e., negotiation protocol) on the market. The Owner n negotiation protocol defines the rules of encounter between Server Infrastructure n Owner Wholesaler of Capacity n trading agents. It should ensure that the negotiation’s likely Wholesaler of Content n outcome satisfies certain social objectives, such as maximizing allocation efficiency (i.e., ensuring that resources are awarded to the participants who value them the most) and achieving market Figure 1. The roles and relationships of stakeholders in the telecom market Virtual enterprise – Information enabler are sold in auctions has grown to tremendous proportions. Content Owner Auctions are defined as a market institution that acts in pursuit of a set of predefined rules in order to compute the desired economic Content Enabler outcome (i.e., high allocation efficiency) of social interactions Server Infrastructure Owner [26]. Based on bids and asks placed by market participants, Wholesaler of Content resource allocation and prices are determined. There are two main PARTNERSHIP FORMATION Virtual enterprise – Transport enabler directions to take when designing auctions, namely we distinguish Carrier efficient and optimal auctions. The objective in efficient auctions BROKERING Network is to maximize allocative efficiency and deal with dividing the Procurament auctions (Multi‐attribute Auction) Infrastructure Owner surplus in an auction among the auctioneer and bidders, while Wholesaler of NEGOTIATION Capacity optimal auctions concentrate on maximizing revenue or the Double‐sided auctions (Continuous Double expected utility of the bid taker [3]. Auction ‐ CDA) BBT model CONTRACT FORMATION 4.1 Multi-attribute auctions Item characteristics (i.e., attributes) represent an important factor CONTRACT in deciding which auction should be used in the negotiation phase. FULFILLMENT Negotiation on commodities, such as transport capacities, focuses SERVICE AND mainly on the price of the item. These items are mostly sold in EVALUATION conventional single-attribute auctions. On the other hand, complex items such as information resources often require negotiation of several attributes, and not just the price [6]. They are sold in multi-attribute auctions [3] which are a special case of Figure 2. The BBT model for the B2B telecom domain procurement auctions. Procurement auctions are also called reverse auctions since there are multiple sellers (e.g., information equilibrium [9]. The negotiation strategy represents a set of rules enablers) and only one buyer (e.g., service provider) that that determines the behavior of a trading agent. purchases items (e.g., information resources). Multi-attribute auctions have been attracting more and more attention in B2B The negotiation process can be either distributive or integrative markets since the price is not the only important attribute [22]. In distributive negotiations, one issue is subject to considered in the decision making process1. negotiation while the parties involved have opposing interests. One party tries to minimize loss and the other party tries to The first step in a multi-attribute auction is for the buyer to maximize gain. Distributive negotiations are also characterized as specify his preferences regarding the item he wishes to purchase. “win-lose” negotiations. The continuous double auction (CDA), Preferences are usually defined in the form of a scoring function which is suitable for transport capacity trading in a B2B e-market, based on the buyer’s utility function [2]. In order to familiarize represents a distributive type of negotiation in a multi-unit auction sellers with buyer’s valuations of relevant attributes, the buyer with multiple buyers and sellers [25]. usually publicly announces his scoring function. Sellers are not obligated to disclose their private values of an item. The winner In integrative negotiations, multiple issues are negotiated while of the multi-attribute auction is the seller that provided the highest the parties involved have different preferences towards these overall utility for the buyer. The buyer sends a request to all issues. For example, two information enablers may want to sell interested sellers which than reply by sending bids. The buyer multimedia information resources to a portal provider, but one is selects the bid with the highest overall utility. If the auction is primarily interested in the sale of news, whereas the other is one-shot, this bid is declared the winning one, otherwise it is interested in the sale of movie clips. These variant valuations can declared as the currently leading bid and the new round of the be exploited to find an agreement resulting in mutual gain. If their auction begins. The buyer can also define the bid increment or preferences are the same across multiple issues, the negotiation minimum requirements the bid has to fulfill in order to compete in remains integrative until opposing interests are identified. In such the next round. Figure 3 shows a multi-attribute auction between a a case, both parties can realize gains: consequently, another name service provider and several information enablers. Information for this class of negotiations is “win-win” negotiations. A multi- enablers offer multimedia content composed of video and audio attribute auction represents an integrative negotiation process streams with different performances. Based on its utility function, which can be used for trading with information resources. the service provider reaches an agreement with the information Last three phases include termination of negotiation where enabler whose information resource has the highest overall utility. negotiated terms are put in a legally binding contract (contract formation), carrying out the transaction agreed in the contract 4.2 Content trading (contract fulfilment) and traders evaluating the received service The term content encompasses movies, songs, news, images and (service evaluation). Due to legal issues and subjective judgments text, in other words data and information within various fields it is not likely that these phases are going to be automated with [14, 23]. The NGN brings its own new added value into the the use of intelligent agents. market and one of these added values is multimedia content composed of several types of content (e.g., audio, video, data…) 4. AUCTIONS 1 Auctions, due to their well defined protocols, are suitable enablers http://www.cindywaxer.com/viewArticle.aspx?artID=149 of negotiations in e-markets. The variety and value of goods that (Business 2.0 magazine) INFORMATION ENABLER e-market, forming value added services from purchased goods (SELLER #1) and then selling those services to consumers on the B2C telecom e-market. Due to the lack of research related to B2B telecom e-markets [13] Business (i.e., most research is related to the B2C telecom e-market [5, 10, 12, 18]) and the expected growth of B2B e-markets in general, the second phase is oriented to finding an appropriate model which INFORMATION AGENT captures all stages related to transactions carried out on the B2B (SELLER ROLE) telecom e-market. As shown in Section 3, the BBT model was used to describe B2B telecom transactions, while intelligent INFORMATION SERVICE PROVIDER ENABLER (SELLER #2) agents were used to impersonate stakeholders on the market. (BUYER) Since the B2B e-market includes repeated transactions with existing and/or new business partners, a new phase should be introduced into the BBT model. This phase will be in charge of collecting knowledge regarding the state of the e-market, Business Business processing information collected in the service evaluation phase, and deciding on the changes that need to be applied in the next INFORMATION AGENT round of negotiations. PROVIDER AGENT (SELLER ROLE) (BUYER ROLE) The third phase is dedicated to the negotiation phase of the BBT telecom model. Well defined and widely researched CDA is used INFORMATION for trading with transport capacities. Consequently, we decided to ENABLER (SELLER #5) focus on multi-attribute auctions for trading with information resources (i.e., content). In order to trade with content, the first step is to define relevant attributes and form an ontology which adequately represents multimedia content. The next step is to Business study existing models of multi-attribute auctions using different approaches (i.e., defining utility functions [2, 3], fuzzy multi- attribute decision making algorithms [24], introducing pricing INFORMATION AGENT functions and preference relations for determining acceptable (SELLER ROLE) offers [1], defining reserved and aspiration levels of attributes and distinguishing negotiable and non-negotiable attributes [4]). Figure 3. An agent mediated e-market for content trading After studying the existing models, we plan to choose the best [11]. When trading with multimedia content there are several features from each approach and try to incorporate them into a attributes that are negotiated on; the quality of the audio and new unified model most suitable for content trading. The new video content (i.e., audio bit rate, resolution of the video), type of approach will be incorporated into to the multi-attribute auction the information provided (i.e., music, video clips, games, news, mechanism based on the English auction. Due to the specifics of sports, weather…), time of origin of the content (e.g., two days the B2B telecom e-market (e.g., larger values of single old weather forecast is of no use, one minute old stock market transactions, repeated transactions, and a smaller number of news could be worth a lot), reusability of the content (i.e., using participants than on B2C e-markets) the goal is to create a balance the content in forming various services), potential number of users between maximizing the allocative efficiency of the B2B market interested in this content, and the price. An example of trading and maximizing revenue or the expected utility of bid takers with multimedia content by using multi-attribute auctions is characteristic for multi-attribute auctions. shown in Figure 3 where several agents posing as sellers offer The fourth phase will be devoted to implementing the multi- different multimedia (i.e., audio and video) content while the attribute auction with agents as representatives of telecom agent posing as a buyer must decide which content holds the stakeholders using the JADE (Java Agent DEvelopment highest utility for him and then buy the content in order to resell it Framework) agent platform and evaluating the designed further on the B2C e-market [18]. mechanism with the existing mechanisms mentioned in the previous paragraph. 5. PROPOSAL OF THE RESEARCH PLAN The research plan consists of four stages2. The aim of the first 6. ACKNOWLEDGMENTS phase is to explore the telecom market, identify participants on The work presented in this paper was carried out within research the market, establish their roles and relationships, and, finally projects 036-0362027-1639 "Content Delivery and Mobility of establish with what goods and services are being traded on the Users and Services in New Generation Networks", supported by market. This phase is completed and is described in Section 2. We the Ministry of Science, Education and Sports of the Republic of can se that a service provider actually manages a supply chain by Croatia, and "Agent-based Service & Telecom Operations buying information and transport capacities on the B2B telecom Management", supported by Ericsson Nikola Tesla, Croatia. 2 I am a 3rd year PhD student. 7. REFERENCES 2007). AAMAS’07. ACM Press, New York, NY, 735-737. [1] Bellosta, M.-J., Kornman, S., and Vanderpooten, D. 2006. DOI= http://doi.acm.org/10.1145/1329125.1329259 An Agent-Based Mechanism for autonomous multiple [14] Hanrahan, H. 2007. Network Convergence: Services, criteria auctions. In Proceedings of the IEEE/WIC/ACM Applications, Transport, and Operations Support. John Wiley International Conference on Intelligent Agent Technology & Sons, Inc. (Hong Kong, December 18-22, 2006). IAT’06. IEEE [15] He, M., Jennings, N.R., and Leung, H. 2003. On Agent- Computer Society, Washington, DC, USA, 587-594. Mediated Electronic Commerce. IEEE Transactions on [2] Bichler, M. 2000. An Experimental Analysis of Multi- Knowledge and Data Engineering. 15, 4 (Jul.y/Aug. 2003), attribute Auctions. Decision Support Systems. 29, 3 (Oct. 985-1003. 2000), 249-268. [16] He, S., Cattelan, R., and Kirovski, D. 2008. Modeling Viral [3] Bichler, M., and Kalagnanam, J. 2005. Configurable Offers Economies for Digital Media. In Proceedings of the 3rd and Winner Determination in Multi-attribute Auctions. ACM SIGOPS/EuroSys European Conference on Computer European Journal of Operational Research. 16, 2 (Jan. 2005), Systems 2008 (Glasgow, Scotland, UK, April 1-4, 2008). 380-394. EuroSys’08. ACM Press, New York, NY, 149-162. DOI= [4] Bui, T., Yen, J., Hu, J., and Sankaran, S. 2001. A Multi- http://doi.acm.org/10.1145/1352592.1352609 Attribute Negotiation Support System with Market Signaling [17] Maes, P., Guttman, R., and Moukas, A. 1999. Agents that for Electronic Markets. Group Decision and Negotiation. 10, buy and sell. Communications of the ACM. 42, 3 (March 6 (Nov. 2001), 515-537. 1999), 81–91. DOI= [5] Damsgaard, J., and Marchegiani, L. 2004. Like Rome a http://doi.acm.org/10.1145/295685.295716 mobile operator's empire wasn't built in a day - a journey [18] V. Podobnik, I. Lovrek, “Multi-Agent System for through the rise and fall of mobile network operators. In Automation of B2C Processes in the Future Internet”, Proceedings of the Sixth Conference on Electronic Student Workshop @ 27th IEEE Infocom, Phoenix, USA, in Commerce (Vancouver, Canada, June 5-8, 2005) ICEC’04. Press. ACM Press, New York, NY, 639-648. DOI= http://doi.acm.org/10.1145/1052220.1052301 [19] Podobnik, V., Petric, A., and Jezic, G. 2008. An Agent- Based Solution for Dynamic Supply Chain Management. [6] Do, V., Halatchev, M., and Neumann, D. 2000. A Context- Journal of Universal Computer Science. 14, 7 (Apr. 2008), based Approach to Support Virtual Enterprises. In 1080-1104. Proceedings of the 33rd Hawaii International Conference on System Sciences (Maui, Hawaii, January 4-7, 2000). HICSS [20] Podobnik, V., Trzec, K., Jezic, G., and Lovrek, I. 2007. 2000). IEEE Computer Society, Island of Maui, 6005. Agent-Based Discovery of Data Resources in Next- Generation Internet: An Auction Approach. In Proceedings [7] Fasli, M. 2007. Agent Technology for E-Commerce. of the 2007 Networking and Electronic Commerce Research Chichester: Wiley & Sons. Conference (Riva Del Garda, Italy, October 18-21, 2007). [8] Fischer & Lorenz (European Telecommunications NAEC 2007. ASTMA, Riva del Garda, 28-51. Consultants). 2000. Internet and the Future Policy [21] Shaw, M. 2000. Electronic commerce: state of the art. In Framework for Telecommunications. Report for the Handbook on Electronic Commerce , M. Shaw, R. Blanning, European Commission T. Strader, A. Whinston, Eds. Springer, Berlin, 3–24. [9] Friedman, D., and Rust, J. 1993. The Double Auction [22] Ströbel, M. 2003. Engineering Electronic Negotiations. Market: Institutions, Theories, and Evidence. Perseus Kluwer Academic/Plenum, New York Publishing, Cambridge. [23] Subramanya, S.R., and Yi, B.K. 2005. Utility Model for On- [10] Furregoni, E., Rangone, A., Renga, F. M., and Valsecchi, M. Demand Digital Content. Computer Magazine. 38, 6 (Jun. 2007. The Mobile Digital Contents Distribution Scenario. In 2005), 95-98. Proc. of the Sixth Int. Conference on the Management of [24] Tong, H., and Zhang, S. 2006. A Fuzzy Multi-Attribute Mobile Business (Toronto, Canada, July 9-11, 2007). IEEE Decision Making Algorithm for Web Services Selection Computer Society, Washington, DC, USA, 32. Based on QoS. In Proceedings of the 2006 IEEE Asia-Pacific [11] Ghys, F., Mampaey, M., Smouts, M., and Vaaraniemi, A. Conference on Services Computing (GuangZhou, China, 2003. 3G Multimedia Network services, Accounting, and December 12-15, 2006). APSCC’06. IEEE Computer user profiles. Artech House, Inc. Norwood, MA, USA Society, Washington, DC, USA, 51-57. [12] Griffin, D., and Pesch, D. 2006. Service Provision for Next [25] Trzec, K., Lovrek, I., and Mikac, B. 2006. Agent Behaviour Generation Mobile Communication Systems – The in Double Auction Electronic Market for Communication Telecommunications Service Exchange. IEEE Transactions Resources. In: KES 2006. LNCS. 4251. Gabrys, B., Howlett, on Network and Service Management. 3, 2. (Second Quarter R.J., Jain, L.C., Eds. pp. Springer, Heidelberg, 318-325. 2006), 2-12. [26] Wurman, P.; Wellman, M.; and Walsh, W. 2002. Specifying Rules for Electronic Auctions. AI Magazine. 23, 3 (Fall [13] Griffin, D., and Pesch, D. 2007. A service oriented 2002), 15-23. marketplace for next generation networks. In Proceedings of the 6th international joint conference on Autonomous agents [27] Yoon, J.-L. 2007. Telco 2.0: A new role and business model. and multiagent systems (Honolulu, Hawaii, May 14-18, IEEE Communications Magazine. 45, 1 (Jan. 2007), 10 -12.