Some Service Issues in Adaptable Service Systems
MetadataVis full innførsel
Networked services have been an important research topic for over 40 years. These days, the amount and variety of services are growing enormously at the same time as the complexity and heterogeneity of the service systems is also increasing. Adaptable services and service systems are a research issue aiming to cope with the complexity. Adaptable Service Systems are service systems that are able to adapt dynamically to changes in time and position related to users, nodes, capabilities, status, changed service requirements and policies. A service can be considered at different abstraction levels. In this thesis, three abstraction levels are used, denoted as the conceptual, engineered and physical services. Service engineering is the creation of conceptual, engineered and physical services. Service management is the functionality to control the provision of service functionality and quality of a service, both within and across service systems, through the service life cycle phases. This thesis addresses some service issues related to service engineering and service management in adaptable service systems. The work presented in this thesis is related with TAPAS (Telematics Architecture for Play-based Adaptable Service Systems). On one hand, TAPAS concepts, architectures and platform are the context and the basis of the thesis. On the other hand, my research work also aims to further develop TAPAS concepts, architectures and platform. The research aims to answer the following five problem statements: P1: How can services be modelled and represented? P2: How can services be discovered efficiently, automatically and accurately? P3: How can services be instantiated dynamically and according to available capabilities and status information? P4: How can new service specifications or modifications to existing services dynamically be introduced without interrupting the executing services? P5: How to evaluate and validate the proposed frameworks and mechanisms? The problem statements P1-P4 are related to the following four research topics: T1: Service representation T2: Service discovery T3: Service instantiation T4: Service adaptation. Service representation is the representation of a service (conceptual, engineered and physical) based on a specific language and a data model. Service discovery is the process of finding services that satisfy functional and non-functional requirements. It is a core functionality to locate desired services in a distributed environment. Service instantiation is the process of creating a service instance upon request and making it available to the user, and finally service adaptation is the process of adapting the structure or behaviour of the service to the various changes during its execution. There is one-to-one mapping from P1-P4 to T1-T4. P5 is related with all the four research topics T1-T4. The problem statements P1 and P2 are further refined into subproblems. The problem statement P1 is refined into sub-problem statements P1 1-P1.3 defined as follows: P1.1: How to represent conceptual services? P1.2: How to represent physical services in a flexible manner so that it is possible to adapt the services to changes dynamically? P1.3: How to extract the component interface behaviour from the physical service representation so that compositional service verification can be applied? The problem statement P2 is refined into sub-problem statements P2 1-P2.2 defined as follows: P2.1: How to ensure automatic and accurate service discovery? P2.2: How to locate services efficiently in a large-scale service system? The result of the research work is classified as nine research contributions C1-C9. These contributions are related to the research topics and accordingly problem statements as defined below: Research topic T1 Service representation: • C1: Conceptual service representation. This contribution addresses P1.1. An integrated semantic service description based on a service ontology is proposed and is represented using Web Services and Semantic Web languages. The service ontology defines a model of functional and non-functional properties, where the service functionality is represented as operations, inputs, outputs, preconditions and effects and the non-functional properties include service parameters, Quality of Service (QoS) parameters and policies consisting of business policies, QoS policies and context policies. Such semantic-annotated service description is the basis for semantic matching procedure in service discovery. • C2: Physical service representation. This contribution addresses P1 2. XML (eXtensible Markup Language) is the physical service representation language. An Extended Finite State Machine (EFSM)-based XML manuscript data model is defined. It is based on modifiable and parameterized behaviour patterns, separating action types from actual action codes. Service functionality is further classified into Action Groups and Capability Categories according to the nature of actions and the dependability on capability respectively. Such manuscript data model is the basis for service instantiation and adaptation. • C3: Preparation for service verification. This contribution addresses P1.3. Service verification is the process of checking service specifications to ensure that service components can play well together. In order to utilize compositional verification based on an interface type language, rules are given for automatic translation from EFSM-based XML manuscript to the interface type language. Projection technique is applied during the translation process. Research topic T2 Service discovery: • C4: Semantic service discovery procedure. This contribution addresses P2.1. An integrated semantic service description model is defined based on a service ontology (i.e. the conceptual service representation). An integrated semantic discovery procedure based on such service descriptions is proposed for semantic matching of both functional and non-functional properties. Such procedure consists of both ontological inference and rule-based reasoning andhas been implemented on a Reasoning Machine (RM). • C5: Super-peer Semantic Overlay Network (SON)-based service discovery system. This contribution addresses P2.2. A service discovery system based on super-peer managed SONs is proposed and functionality for efficient service discovery and efficient SON management is defined. The integrated semantic service discovery procedure proposed for C4 is applied for semantic matching on selected directories (i.e. selected SONs). A self-organizing process based on an autonomous super-peer selection algorithm is applied for super-peer SONs construction and maintenance. The system performance is evaluated by simulations and the results indicate efficient service discovery (in terms of recall, messages-per-request and hops-per-request) and efficient SON management (in terms of self-organization time, management procedure overhead and load factor). Research topic T3 Service instantiation: • C6: Manuscript execution support – State Machine Interpreter (SMI). This contribution addresses P3. This thesis implements an execution support for service instantiation, namely the SMI, which can interpret and execute EFSMbased XML manuscripts. SMI can instantiate the manuscripts according to available capability and status information. Research topic T4 Service adaptation: • C7: Physical service adaptation. This contribution addresses P4. An approach for physical service adaptation is proposed based on the XML manuscripts. Given a service adaptation request, the system dynamically selects and instantiates XML manuscripts according to runtime capability and status information. The actual execution codes for the behaviour patterns defined in the manuscripts can be dynamically selected according to available capability and status. The dynamic generation of such adaptation requests according to traffic situation and failure states is not considered. • C8: Dynamic service management framework. This contribution addresses P4 and is related with C2, C6 and C7. A RM-based framework integrating service behaviour specification (i.e. EFSM-based XML manuscript), selection (instantiation) and adaptation is proposed and prototyped. Selection and Mapping Rules are proposed and modelled. For research topics T1-T4: • C9: Prototypes and simulations. This contribution addresses P5 and is used to evaluate and validate the proposed frameworks and mechanisms.