Multi-Agent Cooperative Design Support in Distributed Environments Camelia Chira1

D. Dumitrescu1

1

Department of Computer Science Babes-Bolyai University, Cluj-Napoca, M. Kogalniceanu 1B, 400084, Romania Tel: +40-264-405300, Fax: +40-264-591906, E-mail: {cchira, ddumitr}@cs.ubbcluj.ro

Abstract The paper addresses system modelling and engineering in cooperative distributed environments. Multi-agent systems, ontologies and web technologies are used for the support and management of resources in a distributed virtual environment. Distributed design involves dispersed heterogeneous teams of people who have to establish and maintain an efficient cooperation process in order to reach global ‘optimal’ solutions to various problems. Key needs addressed by the introduced multi-agent system refer to interoperation among distributed resources and knowledge sharing, reuse and integration. An ontology library creates a common shared understanding of the domain for interoperating agents. The proposed system employs agents for information storage and retrieval, for enhancing collaboration within a distributed design environment and for providing a suitable interface to the user. The agents offer content-related support to the user by committing to the same library of ontologies. The proposed multi-agent design information management system aims to facilitate the optimization of the solution space of the collective dispersed design team.

Keywords: Multi-Agent Systems, Ontologies, Cooperative Distributed Systems, Distributed Design.

1. Introduction Considered an important new direction in software engineering [1, 2], agents and multi-agent systems represent techniques to manage the complexity inherent in software systems and are appropriate to domains in which data, control, expertise and/or resources are inherently distributed [1, 3]. A software agent refers to a computer system situated in an environment which autonomously acts on behalf of its user, has a set of objectives and takes actions in order to accomplish these objectives [2, 4]. The properties that should characterize an agent include autonomy, reactivity, pro-activeness, cooperation, learning and mobility [1-5]. The agents within a multi-agent system must coordinate their activities, negotiate if a conflict occurs and be able to communicate with other agents [1]. Ideal for solving complex problems with multiple solving methods, perspectives and/or problem solving entities, multi-agent systems present many potential advantages including robustness, efficiency, flexibility, adaptivity, scalability, interoperation of multiple existing legacy systems, enhanced speed, reliability and extensibility [1-3] . It is proposed to engage multi-agent systems and ontologies for the support and management of information in a

distributed virtual environment where various physical and human resources dispersed over a computer network have to cooperate to solve different problems. Distributed engineering design is such an environment where heterogeneous teams of people working with different tools and applications have to establish and maintain the processes of communication, coordination, co-location and collaboration in order to determine the global ‘optimal’ solution to the current design problem [6, 7]. A multi-agent design information management and support system is developed to address the key information needs of distributed collaborative engineering design. The structure of the paper is as follows: the needs of the distributed design organization are briefly presented, other approaches to support distributed environments are investigated and the proposed multi-agent and ontological system is designed, implemented and tested.

2. Distributed Design Emerging as a response to market demands and competitive pressures, the distributed engineering design organization involves multiple design teams with heterogeneous skills cooperating together in order to achieve global ‘optima’ in design [6-8]. The human and physical resources involved in the design process can be geographically, temporally, functionally and semantically distributed over the enterprise [5, 7]. Teamwork is playing a significant role in design projects becoming increasingly large, complex and long in duration. Furthermore, the cooperation process among distributed teams of people is crucial for the successful location of the ‘optimal’ design solution [5, 9]. The role of the computer for distributed design is that of a medium facilitating cooperation among distributed designers and also supporting the design process through various applications [9]. Key aspects of this new organization of engineering design that need to be addressed include the support of the cooperation process among participants dispersed across the enterprise and the efficient management of the design related information structures circulated in the distributed design environment [5]. Because the communication of information, the coordination of engineering design participants and team collaboration takes place in a computer based medium, the availability of the software infrastructure to support cooperation and facilitate the management of data, information and knowledge remains the key success factor of distributed design [8]. Advances in the field of Artificial Intelligence (AI) and successful results in other domains (e.g. medicine, commerce) justify the investigation of intelligent problem-

27th International Conference on Distributed Computing Systems Workshops (ICDCSW'07) 0-7695-2838-4/07 $20.00 © 2007

solving methods to support the domain of engineering design. Many of the relevant research studies indicate that the complex activity of distributed cooperative design may be effectively supported by the provision of a collection of interacting autonomous software components incorporating AI specific problem-solving mechanisms [5]. Research shows that AI-based techniques, particularly software agents and multi-agent systems, are a potential successful solution for supporting distributed multidisciplinary design teams collaborating in a virtual environment to achieve global ‘optima’. Started over a decade ago, research in the area of distributed collaborative systems includes a large number of projects which only propose an architecture or framework without providing a viable implementation and testing phase [5]. Difficulties associated with the development of collaborative design systems include the creation of a shared ontology that would enable knowledge-level communication in a distributed environment, the integration of the various design tools and the provision of a cooperation model among interacting participants with different needs and diverse areas of expertise. Furthermore, knowledge management in distributed design needs to be better addressed in order to efficiently facilitate sharing, reuse and integration [10].

3. Proposed Multi-Agent Architecture The proposed Multi-Agent Design Information Management System (MADIS) aims to support the distributed design process by managing information, integrating resources dispersed over a computer network and aiding collaboration processes. It is intended to design a multi-agent system composed of several interacting agent sub-systems in order to deliver these aims. Furthermore, the information circulated within the distributed design environment will be stored in an ontology library to enable content-related support for information management. The main requirements of the proposed MADIS system can be summarized as follows: •

Efficient management of the information circulated in a distributed environment.



Cooperation support through an effective use of communication, co-location, coordination and collaboration processes.



Integration of the heterogeneous software tools used in the distributed environment to enable the flow of information. Figure 1 presents a high-level view over the proposed multi-agent architecture. The Ontological Plane specifies the hierarchy of ontologies that define concepts, relations and inference rules of the engineering design domain. These ontologies compose the machine-enabled framework in which the system’s information resources are circulated and stored. It also includes specific knowledge of the domain instantiated according to the rules specified by the Ontology Library. The scope of the Ontology Library is to create a common

shared understanding of the application domain so that information and knowledge can be shared among the members of the distributed environment. These members can be humans or software agents. The ontology aims to establish a joint terminology between these members [5]. User Agents

Multi-Agent Plane

Interconnection

Application Agents

Ontology Agents

Instance Bases

Ontological Plane

Ontology Library

Figure 1 – Proposed MADIS architecture The Multi-Agent Plane specifies the types and behaviours of the software agents required to enable the system’s functionality. It facilitates the access, retrieval, exchange and presentation of information to distributed teams through various agent systems (e.g. user agents, application agents, ontology agents and interconnection agents). The User Agents form the interface between MADIS and the designer. They provide different services to the user and respond to queries and events initiated by the user (or on behalf of the user) with the help of the ontological agents. The Application Agents have the capability of retrieving information from the applications used by the designer and forward it for storage to the ontological agents. They should be integrated in the software tools regularly used in the distributed engineering design domain and act autonomously pursuing their objective. The Ontology Agents provide ontology management services. They are able to access, retrieve, add, modify and delete information from the ontology library. The Interconnection Agents manage the cooperation process among other agents based on the needs and the services advertised by them. The FIPA [11] agent standard supports the design and development of MADIS. Facilitating agent interoperation, the FIPA agent management ontology is part of each MADIS agent expertise. The agent communication language used in MADIS is FIPA ACL, based on which MADIS agents are able to exchange messages (of types such as request, query, and inform) in order to achieve different objectives. Furthermore, the MADIS ontology completes the expertise of the basic MADIS agents. The implementation phase of the agent societies within MADIS is facilitated by the Java programming language and the Java Agent DEvelopment Framework [12].

27th International Conference on Distributed Computing Systems Workshops (ICDCSW'07) 0-7695-2838-4/07 $20.00 © 2007

4. MADIS Ontology Library The MADIS ontology aims to formally conceptualise the engineering design domain in order to allow knowledge sharing, reuse and integration in a distributed design environment. Creating a semantic link among different architectural components, the ontology library contains specific information ontologies focusing on various aspects of the targeted domain (see Figure 2). MADIS Ontology Library

Shared Distributed Engineering Design

Conceptualization

Explicit Formal Specification

Generic Design Ontology

Detailed Des Ontology

Material Ontology

process, Figure 3 presents the UML-based ontology diagram describing the most important subset of the Structure Ontology. The Material Ontology is used to represent the information regarding the material associated with a part while a Resource Ontology is employed to define the distributed design resources (e.g. human designers, design tools and applications). The Structure Ontology defines the MADIS representation and understanding of a product (considered an important domain entity as the product is the final outcome of the distributed design process). Each product is viewed as a hierarchy of assemblies and parts, with each assembly being made-up of further assemblies (also called subassemblies) and parts. The main constraint defined is that, while a part can be component of an assembly, an assembly cannot be a component of a part. An assembly is considered to be a product if it is not a component of any other assembly. Assemblies and parts are defined in terms of their characteristics (e.g. name, mass, version) and relations (has_author, has_manager, has_feature, has_material) that can link them to instances from other ontologies.

Quality Standards Ontology

Structure Ontology

Design Artefact Ontology

Figure 2 – MADIS Ontology Library A Generic Design Ontology resides at the top-level of the MADIS ontology library. It introduces and defines the main concepts of the distributed design domain e.g. space, time, activity, process and artefact. The other ontologies are specializations of the Generic Design Ontology. Figure 2 depicts some of the ontologies used in MADIS. The Detailed Design Ontology formalizes general concepts specific to the detailed design phase. The Quality Standards Ontology defines the various quality standards and techniques that might be used throughout the design process, so that the artefact will adhere to certain quality standards. The Material Ontology defines concepts and relations about the material properties relating to an artefact (e.g. material type, material ID, ductility, malleability, thermal conductivity and density). The Structure Ontology describes the relationships between the components of the artefact (e.g. fasteners, assembly/disassembly times, routes and tools). Both the Material and Structure Ontologies are specializations of the Detailed Design Ontology. The Design Artefact Ontology is a further specialization of ontologies such as the Material Ontology and the Structure Ontology and will describe the various design parameters of the artefact. The MADIS ontologies are implemented using the Resource Description Framework (RDF) and RDF Schema (RDFS) infrastructure [13]. Furthermore, the development of the MADIS ontology library is supported by the Protégé editor tool [14]. To exemplify the ontology implementation

Figure 3 – The UML diagram of the Structure Ontology

Figure 4 – The Structure Ontology in Protégé Based on the UML diagram created in the design phase, each ontology from the MADIS library has been

27th International Conference on Distributed Computing Systems Workshops (ICDCSW'07) 0-7695-2838-4/07 $20.00 © 2007

implemented using the Protégé editor tool in the RDF/RDFS format. Figure 4 shows the classes of the Structure Ontology represented in Protégé 2000. The ontological instances contained within the MADIS library can be distributed in different locations across the enterprise. These instances are mainly manipulated by the Ontology Agents from the MADIS system.

dynamically generated using Java Servlets supported by the Jakarta Tomcat servlet container and the Apache Web Server. Furthermore, designers can use the MADIS Web Portal to communicate over a computer network based on incorporated tools for instant messaging, video conferencing, application sharing, awareness and whiteboard technologies [5].

5. MADIS Agent Societies The agents specific to each MADIS society have been designed using the AUML concept of Agent Class Diagram, which specifies role(s), state description, actions, methods, capabilities, service description and supported protocols [15]. Furthermore, agent interoperation within MADIS is specified using AUML protocol diagrams are used to model protocols for multi-agent interaction within MADIS. The User Agents implemented in the current MADIS prototype include a User Profile Manager agent (which should act autonomously to manage the profile of the user and should learn user preferences over time) and a User Interface Controller agent (which should provide a customizable graphical user interface based on the user profile). Figure 5 presents the AUML class diagram of the User Interface Controller agent.

-

-

Figure 6 –Search through MADIS Agent GUI

<> USER INTERFACE CONTROLLER (UIC) Organization User Society Capabilities Provides a graphical interface to the user Interfaces various services to access the ontology Protocols Initiates fipa-query protocol with the User Profile Manager for the user profile and MADIS services Responds to fipa-request protocol with the User Profile Manager for updates Initiates a fipa-query protocol with the Ontology Manager for MADIS ontological concepts Initiates fipa-request protocol with the Directory Facilitator for service discovery Initiates fipa-contract-net protocol with the Ontology Broker to ask the provision of different services Collaborators User Profile Manager Ontology Manager Directory Facilitator Ontology Broker

Figure 5 – A MADIS agent class diagram The user can have access to information through a graphical interface (see Figure 6) or through a web-based service (see Figure 7). The information provided to the user in Figure 6 is obtained by the User Interface Controller with the help of the Ontology Agents. These agents are able to read the ontology and forward the information extracted. Alternatively, the user can login to the MADIS Web Portal and have access to the same information in a web page (Figure 7). The MADIS Web Portal offers the functionality of the user dedicated agents in a dynamic web environment. The web pages supplied through the Web Portal are

Figure 7 – Search through MADIS Web Portal The Application Agents are represented by an Application Controller agent integrated in the ProEngineer 2001 CAD application (able to extract information such as part name and mass from the CAD model) and a Component Sender agent (able to forward the extracted information to the Ontology Agents for storage purposes). The integration of MADIS with ProEngineer is realized using a Java toolkit for ProEngineer called J-Link, which allows access to the internal components of a ProEngineer session. The Ontology Agents manipulate RDF/RDFS represented information in order to access, retrieve, add, modify and delete information from the Ontology Library. The Ontology Manager agent supervises the ontology

27th International Conference on Distributed Computing Systems Workshops (ICDCSW'07) 0-7695-2838-4/07 $20.00 © 2007

management process ensuring the consistency of the ontology is accurate and that requested ontology-related services are delivered. The Ontology Broker agent manages the agents that can read the ontology (i.e. the Ontology Reader agents) and the services provided by them. The Ontology Reader agents extract and forward information under certain conditions from the MADIS ontology to the requester agent. The Component Receiver agents have the capability of adding new instances of specific concepts (or components) to the corresponding ontology from the MADIS ontology library. The Ontology Reviser agents can update the ontology by deleting or modifying existing information. The implementation of the Interconnection Agents is fully facilitated by the JADE environment. The System Manager agent supervises the overall functionality of the multi-agent system. All MADIS agents must register with the System Manager in order to allow efficient operation management of the multi-agent system. Based on the FIPA specifications [11], the System Manager must be able to perform functions such as register, deregister, modify, search and get-description. Furthermore, the System Manager has the capability to execute actions such as suspending an agent, terminating an agent, creating an agent, resuming agent execution, invoking an agent, executing an agent and managing resources. The Directory Facilitator agent helps agents to find other agents that provide a requested service. Being FIPA compliant, the Directory Facilitator provides a Yellow Pages service to the agent community. Any agent can use the Directory Facilitator to find other agents providing required services for achieving internal objectives. For example, when a User Interface Controller agent needs to display information regarding a specific concept in a graphical format, the Directory Facilitator can be used to retrieve the agent identifier of the specific Ontology agent(s) that can read the requested information from the Ontology Library. The agent interactions within MADIS are vital for a successful and constructive support provided to the distributed designers. As already indicated, MADIS agents are FIPA compliant and communicate by exchanging ACL messages. The content language used to express the content of the message (communication will be effective if both the sender and the receiver are able to encode and parse expressions using the syntax of the content language) is FIPA SL. Supported by FIPA ACL, the foremost MADIS agent interactions can be summarized as follows: •



The User-Request-Information scenario implies agent interoperation involving the User Interface Controller, the User Profile Manager, the Ontology Broker, the Ontology Reader and the Directory Facilitator agents.

The Application-Save-Information scenario implies agent interoperation among the Application Controller, the Component Sender, the Component Receiver and the Directory Facilitator agents. To exemplify MADIS agent interactions, Figure 8 shows the AUML protocol diagram of the User-Request-

Information scenario. This situation occurs each time the user wants to browse or to search the MADIS ontological instance base. Having the MADIS environment set up on his/her computer, the user can request information through a personal agent managed by the User Interface Controller.

Figure 8 – AUML interaction protocol diagram in MADIS The User-Request-Information scenario involves the following main steps (see Figure 8): (1) the User Interface Controller queries the User Profile Manager for the services provided to the user through a FIPA-QUERY protocol, (2) the User Interface Controller queries the Ontology Manager for the concept categories available in the ontology that can be accessed by the user, (3) the User Interface Controller requests the Directory Facilitator the identification of the agent that can provide the service requested by the user (i.e. Ontology Broker), (4) the User Interface Controller requests the Ontology Broker (identified in the previous step) for the service (e.g. browse, search) needed by the user, and (5) the Ontology Broker instantiates the appropriate Ontology Reader mobile agent that will fulfil the requested service and will migrate back to the User Interface Controller location with the result.

6. Testing and Validation The testing phase of MADIS uses the protocol analysis (PA) technique [16] to evaluate the proposed system when used by a single designer or by a team of designers in a distributed environment to perform a given set of tasks. The subjects were videotaped while using the system and verbalizing their thoughts or communicating with other designers (depending on the task). Besides the MADIS

27th International Conference on Distributed Computing Systems Workshops (ICDCSW'07) 0-7695-2838-4/07 $20.00 © 2007

system, subjects were asked to use traditional groupware technologies to complete similar tasks. The intention was to evaluate the MADIS system itself using the PA approach and furthermore to compare it with groupware technologies currently used by designers (in a best-case real scenario) to share information in a distributed design environment. The groupware technology selected for this reason is Lotus Sametime Document Repository, which allows logged users to manage documents through a web-based interface. The transcripts of each PA session were designed to support the capture and analysis of the subject’s exact verbalization, the observer’s notes and the records of user’s actions [5]. The segmentation of episodes is based on the steps and different screens used by the subject in order to complete the given tasks. Figure 9 shows the episode times for each subject in a PA session where the given tasks refer to the retrieval of specific information about a component in a given assembly. Each subject used the Sametime Document Repository, the MADIS Agents and the MADIS Web Portal to retrieve the requested information. It is clear that the groupware technology was more difficult to be used whereas the MADIS Agents and Web Portal have about the same amount of time allocated. Figure 9 – PA episodes for each subject The PA test results show that agent properties such as autonomy, pro-activeness, cooperation and mobility are highly beneficial to the distributed designer during the information-intensive problem solving process of design. Compared to traditional groupware technologies, the multiagent approach has clear potential benefits including reliability, robustness and faster access to required information (see [5] for more details).

7. Conclusions and Future Work The multi-agent approach to distributed engineering design coupled with the use of ontologies promises to tackle important distributed design issues such as interdisciplinary cooperation among distributed designers, exchange of design data, information and knowledge and integration of heterogeneous software tools. The proposed MADIS system exploits agent’s autonomy, cooperation and learning capabilities in a semantic approach to support a process that involves dispersed heterogeneous resources and multidisciplinary teams. Future work focuses on the extension of the proposed multi-agent and ontological approach to generic distributed environments to facilitate communication and information flows and to manage dispersed resources. It is intended to demonstrate that any cooperative distributed system can be efficiently supported by a framework based on the proposed MADIS multi-agent and ontological model.

Software Engineering: The State of the Art, ed; AgentOriented Software Engineering; ed. P. Ciancarini. M. Wooldridge; AI Volume 1957. [3] Wooldridge, M., Dunne, P. E., 2005. The Complexity of Agent Design Problems: Determinism and History Dependence, Annals of Mathematics and Artificial Intelligence, 45(3-4):343--371. [4] Nwana, H.S., 1996. Software Agents: An Overview, Knowledge Engineering Review, 11(3): 1-40. [5] Chira, C., 2005. The Development of a Multi-Agent Design Information Management and Support System, Ph.D. Thesis., Galway-Mayo Institute of Technology, Ireland. [6] Hirsch, B., 2000. Extended Products in Dynamic Enterprises", E-Business: Key Issues, Applications and Technologies, p. 622-628. [7] Pena-Mora, F., Hussein, K., Vadhavkar, S., and Benjamin, K., 2000. Cairo: A Concurrent Engineering Meeting Environment for Virtual Design Teams, Artificial Intelligence in Engineering, 14(202-219). [8] Chira, O., Chira, C., Tormey, D., Brennan, A., and Roche, T., 2006. An Agent-Based Approach to Knowledge Management in Distributed Design, Special issue on E-Manufacturing and web-based technology for intelligent manufacturing and networked enterprise interoperability, Journal of Intelligent Manufacturing, 17(6). [9] MacGregor, S.P., 2002. New Perspectives for Distributed Design Support, The Journal of Design Research, 2(2). [10] Wang, L., Shen, W., Xie, H., Neelamkavil, J., and Pardasani, A., 2002. Collaborative Conceptual Design State of the Art and Future Trends, Computer Aided Design, 34(981-996). [11] http://www.fipa.org, Foundation for Intelligent Physical Agents. [12] http://jade.cselt.it, Jade. [13] http://www.w3.org, RDF. [14] http://protege.stanford.edu, Protege 2000. [15] Bauer, B., 2001. UML Class Diagrams: Revisited in the Context of Agent-Based Systems, In AgentOriented Software Engineering, Montreal. [16] Ericsson, K.A., and Simon, H.A., 1999. Protocol Analysis: Verbal Reports as Data.

References [1] Jennings, N.R., 2000. On Agent-Based Software Engineering, Artificial Intelligence, 117 (2): 277-296. [2] Wooldridge, M., Ciancarini, P., 2001. Agent-Oriented 27th International Conference on Distributed Computing Systems Workshops (ICDCSW'07) 0-7695-2838-4/07 $20.00 © 2007

Multi-Agent Cooperative Design Support in Distributed ...

and management of resources in a distributed virtual environment. Distributed ... techniques to manage the complexity inherent in software systems and are ...

544KB Sizes 1 Downloads 173 Views

Recommend Documents

Issues in Multiagent Design Systems
Although there is no clear definition of what an agent is .... either a named object (the naming service) or an object ..... design support while leaving room for cre-.

Distributed Vision-Aided Cooperative Localization ... - Semantic Scholar
A similar setup has also been studied in other works, including [5], [9], [10], [11] ...... of a single ground vehicle, equipped with a 207MW Axis network camera8 ..... Proceedings of the International Conference on Field and Service Robotics,.

Graph-Based Distributed Cooperative Navigation ... - Semantic Scholar
Apr 3, 2012 - joint pdf for the case of two-robot measurements (r = 2). ...... In this section, we discuss the effect of process and measurement noise terms on the ..... (50). The computational complexity cost of calculating the .... Figure 5: Schema

Cooperative Cognitive Networks: Optimal, Distributed ...
This paper considers the cooperation between a cognitive system and a primary ... S.H. Song is with Department of Electronic and Computer Engineering, The ...

Part15 - Computer Support Cooperative Work.pdf
Part15 - Com ... ive Work.pdf. Part15 - Comp ... tive Work.pdf. Open. Extract. Open with. Sign In. Main menu. Displaying Part15 - Computer Support Cooperative ...

Part15 - Computer Support Cooperative Work.pdf
Try one of the apps below to open or edit this item. Part15 - Computer Support Cooperative Work.pdf. Part15 - Computer Support Cooperative Work.pdf. Open.

A Software Framework to Support Adaptive Applications in Distributed ...
a tool to allow users to easily develop and run ADAs without ... Parallel Applications (ADA), resource allocation, process deploy- ment ..... ARCHITECTURE.

multiagent systems.pdf
Sign in. Loading… Whoops! There was a problem loading more pages. Retrying... Whoops! There was a problem previewing this document. Retrying.

Distributed Extremum Seeking and Cooperative Control ...
The proposed approach retains all the advantages of cooperative control (such ... Mobile platforms with wireless communication capabili- ties can often be used ...

Distributed Cooperative Q-learning for Power Allocation ...
Since femtocells are installed by the end user, their number and positions are random ..... a femto/macro-user is set to 800 meters, 3) The maximum distance between a .... relatively small overhead. ... Challenges for Business and Technology.

Language Support for Distributed Proxies
lems associated with using proxies in a distributed system. We believe that proxies in ... date) from XML types to Object types and vice versa. We describe these ...

A Distributed Cooperative Power Allocation Method for Campus ...
in power system operation and control. Therefore, building- to-grid integration has recently attracted significant attention. in the smart grid research [3]–[10].

A Distributed Cooperative Power Allocation Method for Campus ...
A Distributed Cooperative Power Allocation Method for Campus Buildings.pdf. A Distributed Cooperative Power Allocation Method for Campus Buildings.pdf.

Collaboration in Distributed Design and Manufacturing using Web ...
Automation, Ohtsu, Japan, July 13-15 1998, (1998). [7] Muammer Koc and Jun Ni, “Introduction of e-Manufacturing”, NAMRC 2003 E-. Manufacturing Panel, McMaster University, May 2003. [8] Wright Paul and Sequin Carlo, “CyberCut: A Networked Manufa

Multiagent Social Learning in Large Repeated Games
same server. ...... Virtual Private Network (VPN) is such an example in which intermediate nodes are centrally managed while private users still make.

Distributed semi-supervised support vector machines
Apr 27, 2016 - The semi-supervised support vector machine (S3VM) is a well-known algorithm for performing semi- supervised inference under ... [email protected] (R. Fierimonte), [email protected]. (P. Di Lorenzo) ... Panella (submitt

Distributed Multicell Beamforming Design Approaching ...
This work is to appear in IEEE Transactions on Wireless Communications. ... Processing Lab, Royal Institute of Technology, Sweden (email: {mats.bengtsson, b-.

organisational change using an online distributed learning support ...
paper introduces the Distributed Learning Support System (DLSS) and the challenges encountered in its ... Later in 1999, a discussion paper was presented to senior management. The paper ... accept a range of file types,. • be browser based,.

Multiagent-Systems-Intelligent-Robotics-And-Autonomous-Agents ...
complete on the web electronic local library that provides usage of large number of PDF file publication collection. You may. find many kinds of e-publication ...

Network Coding in Cooperative Communications ...
S. Sharma is with Computational Science Center, Brookhaven National. Laboratory ... slot, node r forwards the data it overhears in the first time slot to node d.

Distributed Electronic Rights in JavaScript
any powerful references by default; any references it has implicit access to, such as ..... approach solves the open access problem by restricting access to members and regu- ..... Journal of Applied Corporate Finance 8(2), 4–18 (1995). 15.

IN-NETWORK COOPERATIVE SPECTRUM SENSING ...
ber has to be finite, we derive high probability bounds on the iteration ... proposed in-network cooperative spectrum sensing at a given iteration. 1. ... To guarantee high spectrum uti- ..... little impact (logarithmic) on the convergence speed.