Generic autonomic adapter architecture and policy model for semantic socio-cyber-physical collaborative network

Abstract

The cyber-physical system aims to improve the quality of life of citizens by providing intelligent and automated services in a wide variety of sectors like transportations, healthcare,enterprises, self-driving cars, energy sectors and so forth. Recently, considerable amounts of researches have focused on integrating cyber-physical systems in a social context. The idea is to socially connect cyber-physical resources (i.e., physical devices, software elements,networked components, digital contents, etc.) so that they can interact and collaborative for autonomous decision making like humans social networking. However, several challenges remain concerning the designing appropriate methodologies, frameworks and techniques for supporting cyber-physical relation and collaboration within the social context. Most of the existing social software modelling focuses on maintaining human-to-human or human-to object centric interaction only. Existing systems do not recognise how socio-cyber-physical resources can maintain their social status, communicate and interact with both humans and non human entities. The reason may be the lack of understanding and limited approaches or methodologies to semantically (a formal characterisation of the information) represent the socio-cyber-physical resources relation and interactions in a collaborative network. This limits data integration, interoperability, and knowledge discovery from its underlying data sources. Semantic Web’s ontology with a software agent model can help to overcome this limitation by describing and interconnecting socio-cyber-physical objects in a social space.The software agents can act as a representative of these resources to track, manage and update their collaborative activities in a social world.Nevertheless, due to the exponential network growth and uncertainties, the states and relations among socio-cyber-physical objects may keep changing when they are in different situations. Therefore, it is an ardours task and error-prone for humans or traditional software agents to keep track, manage and maintain the larger number of socio-cyber-physical resources and their social dynamics. One potential and flexible solution to this problem is to leverage the autonomic computing approach with social and adaptive goals to make the socio-cyber physical network self-managed and adaptive. Autonomic Computing (AC) approach has laid the necessary foundation to tackle this challenge by developing policy-based Autonomic Adapter (AA) model (e.g., autonomous agent). The AAs can continuously monitor socio cyber-physical resource status, analyse the situation and make a collaborative decision based on the policy knowledge defined by the system administrator. However, autonomic computing model must rely on input knowledge to decide self management operations such as “what”, “where” and “how” to perform the adaptation to the system. Previously, adaptation approaches in a different context have been done in an ad-hoc manner based on the algorithms to predict future circumstances and embed in the program code. This approach is inflexible to dynamic and uncertain environments where system configuration needs to adjust frequently. Defining a flexible policy model and integrating policy into knowledge repository outside the code itself is the most appropriate to manage the autonomic system behaviours during the run-time. Sadly, there has been relatively a little work on developing appropriate policy model and specification language for domain neutral autonomic system.To fulfil the above gaps, our proposed solutions in this thesis has three core contribution to the knowledge. First, we address the establishment of both socio-cyber-physical and human relations and interactions within a social-collaborative network. To achieve this, we propose a software agent-centric Semantic Social-Collaborative Network (SSCN) that provides the functionality to represent and manage cyber-physical resources in a social network. We discuss how nonhuman resources can be represented as socially connected nodes and manage by the software agents. The SSCN is supported by an extended ontology model for semantically describing the concept, properties and relations of human and nonhuman resources. A Java-based software agent API has been implemented to demonstrate some actions performed on behalf of the nonhuman resources in a real-world collaborative healthcare system called, GRiST (www.egrist.org). Second, we propose a Generic Autonomic Social-Collaborative Framework (GASCF) with a policy-based Autonomic Adapter (AA) architecture. The AAs are capable of monitoring system resources, analysing context information, and act accordingly using high-level policy. The AAs can also communicate and exchange data with other AAs through a social network for collaborative decisions making like human social interaction.Third, we propose Event-Condition-Action (ECA) rule-based policy model and specification language for AA by defining Policy Schema Definition (PSD) and Policy Script Specification(PSS) languages, modelled with XML syntax. Finally, we test and evaluate our approach by implementing it to the extended GRiST socio-healthcare service context and eGRiST clinical decision support system. We demonstrate and evaluate how socio-cyber-physical relation,interaction and autonomous decision-making is achieved by integrating AAs and using policy specification to manage AAs behaviour within socio-cyber-physical medical context.