Assessing Dynamic Service Discovery In The Network Centric Battlefield

Document Sample
Assessing Dynamic Service Discovery In The Network Centric Battlefield Powered By Docstoc
					                            Projet de synthèse bibliographique

                                     Fiche de lecture
Key-words :
Service discovery, Network Centric Warfare (NCW), Network Centric Battlefield, pervasive, registry
network, topology, service-oriented architecture, Registry bootstrapping, dynamic environment


Lector : Julien BARBOT


Diffusion : Département INFORMATIQUE
             INSTITUT NATIONAL DES SCIENCES APPLIQUEES DE LYON


Document :
    -   Type : Research paper
    -   Title : “ASSESSING DYNAMIC SERVICE DISCOVERY IN THE NETWORK CENTRIC BATTLEFIELD”
    -   Authors : Tommy Gagnes
    -   Published by: Norwegian Defence Research Establishment (FFI)
    -   Source: IEEE Xplore


Summary :
This document is focused on service discovery architecture. Particularly on dynamic environment
such as in Network Centric Battlefield or Network Centric Warfare (NCW). The following topics are
discussed:
    - Requirements for dynamic service discovery, as Web-Service technologies are not sufficient.
    - Design aspects of service discovery architecture, which emphasis autonomous federated
        registry topology

Part 1 – Introduction / Issues
    1) Concept of dynamic service discovery - to connect a producer to a consumer node
        dynamically (ex. In Network Centric Battlefield opportunistic and rapid discovery of services
        can enhance the amount and the accuracy of information accessible to war fighters).
    2) Concept of registry – A registry node is used to connect a consumer to a producer
        dynamically. It concentrate some services available from producer (with the meta-data
        associated) and help consumers to find the appropriate services.
    3) Problem of topology in dynamic environment – Dynamic environment lead to frequent
        change of service meta-data and topology (service producer nodes & registry nodes).
    4) Problem of configuration & maintenance – Service discovery should reduce manual
        configuration and enable automatic discovery and selection of relevant services. It should
        also be robust and bandwidth efficient.

Part 2 – Requirements for dynamic service discovery
    1) Requirements of bootstrap – Unified way for automatic discovery of registry nodes to avoid
        manual configuration and maintenance.
    2) Requirements of delegation – Service selection is done by registry nodes, to allow flexible
        resource utilization (limit the amount of bandwidth and CPU time used by consumer nodes).
    3) Requirement of robustness – Cannot depend on a centralized components (such as a
        registry), to prevent system failure.
    4) Requirements of heterogeneous service description – Simple (URI) or Rich (semantic,
                            Projet de synthèse bibliographique

       ontology, web …).
    5) Security (cf. 2nd problematic on authentication)

Part 3 – Aspects of a service discovery architecture
     Service description (cf. COCOA paper for more information) – Description of Name, type,
        operations, parameters (input & output), attributes. Goals and processes can also be
        modeled.
          - Simple description, string or URI. Typically the case for Web Services, see WSDL and UDDI
          - Richer description, based on semantic. Allow to describe whether a service can perform a
            task. Allow to reduce ambiguity, and to dynamically use services (cf. Service composition,
            COCOA), see OWL-S , WSMO, and SAWSDL
     Registry technologies – Possess their own information model, maintain a list of services
        available. Allow consumer to perform queries and lookup. The service list can be indexed
        upon a hierarchy or a taxonomy. See ebXML.
     Registry network topology – a)Centralized, b)
        Decentralized, c) Hybrid. The c) is a good compromise
        with technology based on distributed hash table.

       Registry network bootstrapping and maintenance – LAN & WAN are considered separately.
        To find out about registry node, registry discovery must be carried out.
         - On LAN, whether by multicast or registry nodes can periodically send local beacon
           messages.
         - On WAN, multicast is too bandwidth consuming. Seeding or P2P system should be use.
         - Once a registry node is found (which is connected to another registry nodes), alternative
           registry can be provided to client to prevent failure.
       Maintaining service information in registry – Registry should mirror the state of available
        services, as the topology evolve quickly obsolete services have to be removed from the
        system.
         - Concept of lease – A provider should periodically confirm it’s aliveness.

Part 4 – Towards a coherent architecture for service advertisement distribution
     Service architecture triangle (see figure 2),
        client nodes, registry nodes, and service
        nodes
     System        based     on      autonomous,
        dynamically federated registries provide
        the best balance.
     Several service description and query
        mechanisms should coexist. A layered
        approach is described (see figure 3)
                           Projet de synthèse bibliographique

Interest of this document for the study
A lot of information about service discovery, many references to another works and well described.
Limitation about Web Services technologies for pervasive and dynamic environment. Proposal of
new and pertinent architecture.
Requirements for dynamic service discovery.

Contains many information we could reuse on our synthesis, and many links we should explore.

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:4
posted:3/30/2012
language:
pages:3