Current Developments and Future Challenges of Coordination in Pervasive Environments Manfred Bortenschlager Salzburg Research Jakob Haringer Straße 5/III, A-5020 Salzburg, Austria
[email protected] Gabriele Kotsis Johannes Kepler University Linz Altenbergerstraße 69, A-4040 Linz, Austria
[email protected]
Abstract With the miniaturization of computing elements, their increasing number, interconnection, and penetration of our daily lives, pervasive computing environments start to become more and more complex. This results in an apparent claim for improved coordination mechanisms, to optimize, to enhance effectiveness and eventually to provide the users with a sufficient level of quality. We took this as our motivation to investigate the concept of coordination applied to pervasive environments in more details. This report is the summary of the CoMA workshop held at IEEE WETICE 2007 and highlights the current research efforts and outlines potential future challenges with respect to coordination in pervasive environments.
1
Motivation
Recent developments and technological advances in information and communication technology (ICT) are leading to an increasing availability and functionality of mobile, portable, and embedded devices, and an improved qualityof-service of wireless connections together with decreasing costs [7]. The Internet and (mobile) telecommunication networks are currently converging to one vast and ubiquitously available information space. Consequently, a great magnitude and diversity of entities involved in such pervasive environments are emerging. We are moving towards the “Internet of Things” [4] and we are shifting from the one person is associated with one computer paradigm to a oneto-numerous relationship. In addition, these trends and technologies shall be
Marco Mamei University of Modena and Reggio Emilia Via Allegri 13, Reggio Emilia, Italy
[email protected]
exploited to beneficially assist the users’ in their daily activities—but as unobtrusively as possible; as it is the major claim and challenge at the same time of pervasive computing [8]. Mark Weiser described this most accurately in his article [10] as “The most profound technologies are those that [...] weave themselves into the fabric of everyday life until they are indistinguishable from it”. People and their collaborative behaviors shall be supported by using pervasive computing technology. We argue that an appropriate engineering of pervasive computing environments with an explicit handling of coordination mechanisms is inevitable in order to eventually provide the conveniences of ubiquitous services delivered with an acceptable quality. These considerations represent our motivation for the organization and conduction of the workshop about Interdisciplinary Aspects of Coordination Applied to Pervasive Environments: Models and Applications (CoMA)1 held at the 16th IEEE International Workshops on Enabling Technologies: Infrastructures for Collaborative Enterprises (WETICE)2 from June 18-20, 2007 in Paris, France. In this report, we give a brief introduction to coordination (Section 2), summarize the technical issues presented at the workshop (Section 3), highlight the topics discussed (Section 4), show an excerpt of coordination concretely applied to European projects (Section 5), and conclude with a summary and a proposal for future research challenges (Section 6).
1 See 2 See
http://mowi.salzburgresearch.at/wetice http://www.wetice.org
2
Introduction to Coordination
The characteristics of pervasive and mobile environments like unpredictable, sporadic connectivity, or high dynamics with respect to topology changes, quality of service, and frequently changing data result in particular requirements for coordination mechanisms, which shall be the prime concern of research activities conducted with respect to engineering coordination in pervasive environments. Coordination is significantly responsible for the effectiveness, performance and quality of complex systems. An explicit handling and a careful integration of accurate coordination mechanisms enables an enhancement of the overall system behavior and contributes to improvements in terms of flexibility, adaptability, maintenance, unobtrusiveness, and service quality. The concept is cross-sectional and a great impact may be reached in different application areas if effective methods and models can be designed, implemented, validated, and deployed. Coordination as a concept is quite abstract and inter- and multidisciplinary. Nevertheless, people have a good intuitive notion about what coordination actually means. It penetrates our daily lives and is omnipresent; and mostly becomes apparent when it is lacking or badly “implemented”. The same or similar strategies or mechanisms may be applied to very diverse disciplines [6] such as computer science, social sciences, anthropology, biology, finance, sociology, psychology etc. and it may appear in many different facets such as direct or indirect coordination, collective or individual, intentional or unintentional, competitive or benevolent, centralized or decentralized, a priori defined or ad hoc determined, or under authoritarian control or autonomously conducted. However, the principal task of coordination is always to achieve a common, superior goal as effectively, fast, and economically as possible. In [9], coordination is said to be the “integration and harmonious adjustment of individual work efforts towards the accomplishment of a larger goal”. The interdisciplinary study of coordination (i.e. coordination theory) [6] tries to tackle this issue and to formalize this abstract concept of coordination. Malone and Crowston used a more formal approach to describe coordination theory as an interdisciplinary subject which comprises four entities [6]: goals, activities, actors, and interdependencies. This theory defines coordination as the “act of managing interdependencies between activities”. Consequently, the challenge shall be to investigate and to elaborate accordant mechanisms to resolve the interdependencies, which is the key to improve coordination. To accomplish this, coordination laws are established and applied defining how to resolve the interdependencies. Proposals for architectures or models for coordinative systems are sporadically available such as in [2] and [5]. Ciancarini states that coordinative processes can always be
described by the triple (i.e., the coordination triple) of {E, M, L} [2]. In this generic model, {E} represents the coordinable (either physical or logical) entities which have to be coordinated. These can be (software) processes, services, (software) agents, or even human beings. {M} stands for the coordination media (i.e. communication channels), which serve as connectors between the entities and facilitate communication. Instances of coordination media may be a message-passing environment, remote procedure calls, or shared spaces. {L} is referred to the coordination laws defining how the interdependencies which are the focal point in coordination have to be resolved and hence, semantically define the coordination mechanisms.
3
The CoMA Workshop
The workshop particularly focused on discussions about coordination and pervasiveness, the current state of research efforts, and practical experiences with the clear vision of generating mutual benefits for the participants. To support this, opportunities for showcases, demonstrations, and indepth discussions were provided. CoMA was divided into four sessions where the fourth one was exclusively used for intensive discussions on the topic. The following eight full and two short papers were accepted and presented at CoMA which can basically be categorized according to the following three types: (i) Pervasive Coordination Middleware, (ii) Context-dependent Coordination, or (iii) Coordination in Pervasive Multi-agent Systems: 1. Coordination in the Sensor’s Continuum by Nicola Bicocchi, Gabriella Castelli, Alberto Rosi, and Franco Zambonelli This paper introduced an infrastructural approach to exploit service invocation based on a “sensor’s continuum” with a particular focus on modeling context data (the W4 model). With this, the authors pursue the Ubiquitously Browsing the World vision [1] which shall be enabled by the provision of context-aware coordination of ubiquitous services. 2. A Bayesian Approach for Disconnection Management by Massimiliano de Leoni, Massimo Mecella, and Ruggero Russo As the communication medium is an essential prerequisite for coordination, this paper deals with an algorithm to avoid imminent disconnections between entities connected through a MANET configuration. A predictive technique to estimate possible disconnections in such mobile scenarios was proposed and empirically evaluated.
3. XVSMP/Bayeux: A Protocol for Scalable Space Based Computing in the Web by eva K¨uhn, Johannes Riemer, and Lukas Lechner This paper proposes to overcome shortcomings of present web infrastructures and architectures—such as inflexibility and the inability of server-side push information delivery—by integrating space-based computing (SBC) technologies. By this, the inherent decoupling mechanisms of SBC can be exploited where the HTTP protocol is used to transport space operations providing improved behavior of Web-based information delivery. 4. Coordinating Knowledge in Pervasive Environments by Lyndon Nixon, Robert Tolksdorf, Alan Wood, and Ronaldo Menezes The authors introduce the Semantic Web Spaces prototype, which represents a new coordination model and API based on Linda and tuple spaces. They extend the classic tuple space approach by arguing for the coordination of knowledge. Novel concepts are required to solve problems of knowledge coordination. This work contributes a new API including the use of scopes and fading to provide a measure for the relative strength of knowledge claims in the system (i.e., how much clients can “believe”). 5. Enabling Collaborative eHealth Through Triplespace Computing by Lyndon Nixon, Elena Simperl, Dario Cerizza, Emanuele Della Valle, and Reto Krummenacher Triplespace computing is a new semantic tuple space paradigm which shall revolutionize machine-based communication just as the Web’s persistent publication paradigm and simple interaction pattern revolutionized human communication. To make Triplespace computing a reality, the authors align communication models (i.e., Linda and tuple spaces) to the knowledge-centered approach (i.e., RDF and reasoning). A decentralized architecture supports emergent coordination of intelligent agents or services. The approach is applied to the eHealth domain, where knowledge is represented as RDF graphs. 6. Laws for Mediating Agents’ Activities in Situated Multiagent Systems by Danny Weyns and Tom Holvoet This paper focuses on the importance of engineering the environment in which agents are embedded. The authors argue that for many particular applications it is more beneficial to encapsulate the coordination logic within the environment rather than in the coordinables (i.e., agents in this case). Three types of laws integrated in the environment are described: perception, action, and communication laws.
7. Building Smart Environments as Agent Workspaces by Andrea Omicini, Alessandro Ricci, and Giuseppe Vizzari The paper describes an approach based on agents and artifacts used to model and design pervasive environments. The authors refer to “Smart Environments” that are specific work areas supporting the interaction among actors, and between actors and resources in order to facilitate work activities. This work tries to stress the relevance of social aspects that are sometimes overlooked in pervasive research. 8. WSDir: a Federated Directory System of Semantic Web Services by Michael Schumacher, Tim van Pelt, Ion Constantinescu, Alexandre de Oliveira e Sousa, and Boi Faltings In this work, the design and implementation of the distributed directory for semantic (Web) services (WSDir) is described. The typical use of this directory system is in a context where ubiquitous business services should be flexibly coordinated and provided to the mobile user in dynamically changing environments. WSDir has been applied and evaluated as a backbone in the trials of an eHealth emergency application where agents are in charge of Web service coordination. 9. Analyzing and Modelling Office Activities (short paper) by Michele Bezzi, Robin Groenevelt, and Frederick Schlereth The authors propose an agent based model to describe people movements and derived activities and provide experimental data by integrating various types of sensors (in an office environment). From this, simple interaction or collaboration rules can be reproduced from the observed structures and behaviors which can, for instance, be exploited in terms of according office layout. 10. SCOMET: Adapting collaborative working environments to the MANET scenario (short paper) by Marcel Arrufat, Gerard Paris, Pedro Garcia Lopez, and Antonio F. Gomez Skarmeta SCOMET represents a peer-to-peer based middleware for collaborative applications building up on MANET environments. This middleware can transparently be deployed over several network configurations and offers services to higher layers such as publish/subscribe or group management, which can be exploited for CSCW applications, for instance.
4
Discussion
The entire last session of CoMA was devoted to discussions only, where each paper was briefly mentioned again containing a one statement summary, its contribution to coordination, and future challenges with respect to coordination in pervasive environments. Additionally, everyone again got the chance to criticize or defend the described idea. We give a synthesis of the main issues arising in this section. As tuple spaces-based approaches were one of the prevailing topics, once again this paradigm seems to be highly appropriate for exploiting it to realize coordinative systems. Its simplicity and diversified functionality at the same time, make it again—although this idea has been around now for more than 20 years, see [3]—a very attractive underlying architectural approach to implement (coordination) middleware for pervasive environments. A further important issue was that—although there is an increasing need for coordination in pervasive environments apparent—in order to provide the respective coordination mechanisms, we need to better understand the specific and real requirements and characteristics of such environments. The difference to “conventional” coordination systems addressing stationary applications has to be treated explicitly. Especially the high degree of dynamics has to be tackled accordingly. To date, unfortunately, coordination system engineers do not appropriately understand the real user needs with respect to pervasive environments. Hence, more user-centered approaches would be advisable—accomplished in iterations with user-feedback opportunities. The users should not be left out of these research and development loops. As the environments become more and more pervasive and “intelligent” a proper consideration or engineering of these is necessary. The question is where to put the coordination “logics” or laws; either encapsulate them in the coordinables (i.e., entities) or in the environment. This remains an open (research) question and highly depends on the application and its requirements. Furthermore, it was argued that not the adopted middleware technology might be decisive but understanding the context and providing simple and expressive ways to model and manage context. The coordination community spent a lot of effort in examining coordination based on data. With the uptake of the Semantic Web, however, other means for representing knowledge become available. Hence, we may face a beneficial shift to coordination based on knowledge. Attention will have to be paid on a more thorough consideration, integration, and exploitation of semantics in coordination systems. It should be particularly focused on engineering knowledge and on ontology engineering. On a broader basis, which presumably applies for com-
puter science more generally, re-inventing the wheel should be avoided by more thoroughly investigating related works. Additionally, due to the fact that coordination in pervasive environments is a rather young field of research, many prototypical implementations may lack scalability and robustness which both are characteristics of prime importance for a system in order to get accepted on the long run.
5
Application of Discussed Research Works
Most of the works presented and discussed during CoMA are concretely applied in European research projects. This section represents an excerpt and summarizes the coordination related research works with respect to selected projects. CASCADAS CASCADAS stands for “Component-ware for Autonomic Situation-aware Communications, and Dynamically Adaptable Services”. Facing a tremendous increase in available ICT resources, users shall be supported in configuration issues of the arising complex interrelations of entities. The goal is to demonstrate an innovative architectural vision based on self-organized distributed components for the provision of autonomic and situation-aware communication services3 . Among others, paper 1 is relevant for and applied to this project. CASCOM The main objective of the project CASCOM (Contextaware Business Application Service Co-ordination in Mobile Computing Environments) is to implement, validate, and trial a value-added supportive infrastructure for Semantic Web based business application services across mobile and fixed networks. The essential approach of CASCOM is the innovative inter-disciplinary combination of intelligent agent, Semantic Web, P2P, and mobile computing technology. The health care use case is demonstrated4 . Among others, paper 8 is relevant for and applied to this project. POPEYE The main goal of the research project POPEYE (“Peer-toPeer Collaborative Work Environments over Mobile AdHoc Networks”) is to provide the concepts, methods and core services for next generation mobile collaborative working environment with emphasis on P2P information exchange model in the environment of heterogeneous mobile 3 See 4 See
http://www.cascadas-project.org http://www.ist-cascom.org
ad hoc networks5 . Among others, paper 10 is relevant for and applied to this project. TripCom TripCom stands for Triple Space-based computing and is based on the evolution and integration of several wellknown technologies such as Tuple Space Computing, Shared Object Space, Persistent Message-based Architecture, and RDF. The goal is to integrate Tuple Spaces, Semantic Web (triple), and Web service technologies. For this, Tuple Space technology shall be improved by adding semantics and means to structure and relate tuples in a scalable and linkable Triple Space architecture. Web service technologies shall be extended by adopting the flexible and powerful asynchronous communication model of Tuple Spaces6 . Among others, papers 3, 4, and 5 are relevant for and applied to this project. WORKPAD The WORKPAD project aims at developing an innovative software infrastructure for supporting collaborative work of human operators in emergency scenarios. In such scenarios, different teams and team members shall be coordinated by a context-aware and P2P-based infrastructure that adaptively monitors and manages the workflows always relative to the current situation on site. The teams are provided with context-dependent information coming from a backend grid infrastructure that semantically integrates various data sources7 . Among others, paper 2 is relevant for and applied to this project.
6
Summary and Future Challenges
The prime focus of the CoMA workshop was coordination in pervasive environments where basically three broader topics were discussed: (i) Pervasive Coordination Middleware, (ii) Context-dependent Coordination, and (iii) Coordination in Pervasive Multi-agent Systems. Discussions were very lively and fruitful and some potentially relevant future research challenges were identified: - Better understanding and tackling the specific requirements of pervasive environments
- Reasoning about where to encapsulate the coordination logic - Integrating and exploiting semantics in coordination systems (ontology engineering) - Adopting more user-centered approaches Some presented works, in fact, are complementary and collaborations seem to be fruitful. An ideal effect of this workshop would be fostering joint (research) initiatives, reasoning about the discussed issues and problems, generating new ideas, and further contributions to improve coordination in pervasive environments. Acknowledgements. The accomplishment of this workshop is supported by the European Commission through the FP6-2005-IST-5-034749 project WORKPAD8 .
References [1] G. Castelli, A. Rosi, M. Mamei, and F. Zambonelli. Browsing the World: Bridging Pervasive Computing and the Web. In International Workshop on Ubiquitous Information Systems, Muenster, Germany, 2006. [2] P. Ciancarini. Coordination Models and Languages as Software Integrators. ACM Comput. Surv., 28(2):300–302, 1996. [3] D. Gelernter. Generative Communication in Linda. ACM Trans. Program. Lang. Syst., 7(1):80–112, 1985. [4] N. Gershenfeld, R. Krikorian, and D. Cohen. The Internet of Things. Scientific American, 291(4):76–81, October 2004. [5] M. Klein. Coordination science: Challenges and directions. In Coordination Technology for Collaborative Applications, pages 161–176, 1996. [6] T. W. Malone and K. Crowston. The Interdisciplinary Study of Coordination. ACM Comput. Surv., 26(1):87–119, 1994. [7] F. Mattern and P. Sturm. From Distributed Systems to Ubiquitous Computing – The State of the Art, Trends, and Prospects of Future Networked Systems. In K. Irmscher and K.-P. F¨ahnrich, editors, Proc. KIVS 2003, pages 3–25, Springer-Verlag, Feb. 2003. [8] M. Satyanarayanan. Pervasive Computing: Vision and Challenges. IEEE Personal Communications, pages 10–17, Aug. 2001. [9] B. Singh. Invited Talk on Coordination Systems, 1989. Organizational Computing Conference, November 13-14, Austin, Texas. [10] M. Weiser. The Computer for the 21st Century. Scientific American, 265(3):66–75, Sept. 1991.
- Supporting and exploiting the shift from data-based coordination to knowledge-based coordination - Developing expressive concepts for modeling context to allow for context-dependent coordination 5 See
http://www.ist-popeye.org http://www.tripcom.org 7 See http://www.workpad-project.eu 6 See
8 See
http://www.workpad-project.eu
