Achieving Autonomicity in IoT systems via Situational-Aware, Cognitive and Social Things Orfefs Voutyras, Spyridon Gogouvitis, Achilleas Marinakis and Theodora Varvarigou, National Technical University of Athens Presenter: Orfefs Voutyras 18 th Panhellenic Conference on Informatics (PCI 2014) October , Athens, Greece

03/10/ Overview Goal Introduction The concept of Knowledge The concept of Experience Types of learning Learning through communication Social properties of the Virtual Entities Management components Summary

03/10/ Goal Our aim is to support knowledge flow between Things in order to provide a system that acts in an autonomous way, learns, observes and evaluates the usage and communication patterns and generates new knowledge. Our proposal focuses on the value of experience and experience-sharing and investigates models and principles designed for the social networks, which would provide it with the potential to support novel applications in more effective and efficient ways.

Introduction (1/3) The COSMOS project will provide a framework for the decentralized and autonomous management of Things based on service-, interaction-, location- and reputation-oriented principles, inspired by social media technologies. Achieving Autonomicity via Situational-Aware, Cognitive and Social Things 03/10/2014 4

5 Introduction (2/3) The proposed approach follows the: IoT-A reference model –Virtual Entities (VEs) and Groups of Virtual Entities (GVEs). Social Internet of Things (SIoT) paradigm – it maps the social relations and interactions of the individuals to their VEs. – it defines, monitors and exploits social relations and interactions between the VEs. – it uses technologies and exploits services from the domain of the social media. MAPE-K model – self-management and – autonomicity

03/10/ The COSMOS MAPE-K loop CognitiveSocial Situational- aware A P SA SM E M Knowledge Component Introduction (3/3) Monitor-Analyze-Plan-Execute (MAPE): an autonomic control loop or autonomic manager as proposed by IBM. In addition to the MAPE components, an autonomic manager also contains a Knowledge block that is connected to all four of the MAPE functional blocks, producing a MAPE-K control loop. We extend the MAPE-K loop by introducing two new components, Social Monitoring (SM) and Social Analysis (SA).

03/10/ The concept of Knowledge. know-Best know-How know-What know-Nothing learning planning analysis monitoring Data Information Knowledge Wisdom raw-data collected through IoT-services The COSMOS DIKW Pyramid

03/10/ The concept of Experience (1/2) Experience can be: a piece of Knowledge described by an ontology, a Model resulting from Machine Learning or contextual information We focus mainly on the representation of experience through Cases as defined in the Case Based Reasoning (CBR) technique. A case can be considered as a combination of a problem with its solution, whereas a problem consists of one or more events. In other words, a case is a kind of rule for an actuation plan, which is triggered when specific events are identified.

03/10/ The concept of Experience (2/2) Ontologies are used for the description of the VEs Cases are one form of Experience Used as a means to reason: cause and effect (Problem – Solution) Each VE may maintain its own Case Base (CB) locally as part of its KB Classes of the COSMOS ontology

03/10/ Types of learning Individual Learning, through self enrichment of local CB. Learning through communication, by using the experience sharing (XP-sharing) mechanism. Learning through a knowledge repository, when the VE connects to the COSMOS platform.

03/10/ Individual Learning A general CBR cycle may be described by the following four processes: RETRIEVE the most similar case or cases REUSE the information and knowledge in that case to solve the problem REVISE the proposed solution RETAIN the parts of this experience likely to be useful for future problem solving The CBR cycle (adapted from [Aamodt, 1994])

03/10/ Learning through communication Forms of learning through communication: – Demand driven or – Supply driven learning. Influences overhead and hit rate. Dissemination options – Broadcasting – Narrow casting – Personal casting Learning through communication

03/10/ Social properties of the virtual entities Concept of Friends that act more like Twitter’s Followers. Used for greater versatility of communication (decentralization) and knowledge acquisition. Choice based on Relevance and Dependability. Relevance includes VE Domain and Physical Entity matching (Homophily), as well as Distance proximity through Location and Geo-location measurements. Dependability measures social willingness and usefulness of shared knowledge (Trust, Reputation), as well as absence of mechanical failures (Reliability). Example of VEs’ properties

03/10/ Management components Profiling and Policy Management (PPM) for assigning Unique VE IDs and maintaining the openness factors of individual VEs. Friends Management (FM) for creating and maintaining the Friend List, as well as providing suggestions to the user. Social Monitoring (SM) in order to evaluate feedback on all social actions concerning the VE. Social Analysis (SA) so that the platform can retrieve data from VEs’ SM components and extract complex social characteristics of the VEs as well as models and patterns in intra VE communication.

03/10/ Summary The COSMOS platform can be characterized as a SIoT platform since it defines, monitors and exploits social relations and interactions between the VEs and uses technologies from the domain of the social media. The social side of COSMOS improves the knowledge flow (distributed knowledge) and introduces the concept of experience sharing between Things, enabling Things to react in a more autonomous way. However, one of the main concerns regarding the success of such an architecture is its potential to maintain an opportunistic IoT system, offering the human users motives to share the knowledge and IoT- services of their VEs.

Orfefs Voutyras NTUA Further Information: The research leading to these results is partially supported by the European Community’s Seventh Framework Programme under grant agreement n° , in the context of the COSMOS Project. Thank you!

Managed system: The system collects and offers to the administrator all the information needed to take decisions. Predictive system: The system is able to recognise patterns, predict the optimal configuration and make proposals to the administrator. Adaptive system: The system is able, not only to “offer advice” for certain actions, but can trigger on its own the right actions, based on the information that it has gathered. Autonomic system (real autonomy): The system’s actions are based on business rules, models and goals. The users react with the system only when some changes to these rules are needed. 18 Autonomous Management