Benoit Gaudin

Knowledge representation for self-adaptive behavior

An autonomic system is considered to be a self-adaptive system that changes its behavior in response to stimuli from its execution and operational environment. Such behavior is considered autonomic and self-adaptive and is intended to drive intelligent systems in situations requiring adaptation. Such systems encapsulate rules, constraints and mechanisms for self-adaptation and acquire and process knowledge about themselves and their environment. In this paper, an approach to knowledge representation and reasoning for self-adaptive behavior is presented. The approach is formal and demonstrates how knowledge representation and reasoning help to establish the vital connection between knowledge, perception, and actions realizing the self-adaptive behavior. The knowledge is used against the perception of the world to generate appropriate actions in compliance to some goals and beliefs.

A control theory based approach for self-healing of un-handled runtime exceptions

This work presents an approach to self-healing that deals with un-handled exceptions within an executing program. More precisely, we propose an approach based on control theory that automatically disables system functionalities that have led to runtime exceptions. This approach requires the system to be instrumented prior to deployment so that it can later interact with a supervisor. This supervisor encodes the only sequences of actions (method calls) of the system that are permitted. We describe an implementation that automatically generates instrumentation for Java systems and demonstrate the efficacy of this approach through a comprehensive example.

Software Maintenance through Supervisory Control

This work considers the case of system maintenance where systems are already deployed and for which some faults or security issues were not detected during the testing phase. We propose an approach based on control theory that allows for automatic generation of maintenance fixes. This approach disables faulty or vulnerable system functionalities and requires to instrument the system before deployment so that it can later be monitored and interact with a supervisor at runtime. This supervisor ensures some property designed after deployment in order to avoid future executions of faulty or vulnerable system functionalities. This property corresponds to a set of safe behaviors described as a Finite State Machine. The computation of supervisors can be performed automatically, relying on a sound Supervisory Control Theory. We first introduce some basic notions of Supervisory Control theory, then we present and illustrate our approach which also relies on automatic models extraction and instrumentation.

Requirements and initial model for KnowLang: a language for knowledge representation in autonomic service-component ensembles

Autonomic Service-Component Ensembles (ASCENS) is a class of multi-agent systems formed as mobile, intelligent and open-ended swarms of special autonomic service components capable of local and distributed reasoning. Such components encapsulate rules, constraints and mechanisms for self-adaptation and acquire and process knowledge about themselves, other service components and their environment. ASCENS systems pose distinct challenges for knowledge representation languages. In this paper, we present requirements and an initial model for such a language called KnowLang. KnowLang is intended to provide for formal specification of distinct knowledge models each representing a different knowledge domain of an ASCENS system, such as the internal world of a service component, the world of a service-component ensemble, the surrounding external world and information of special situations related to state changes and operations of service components. KnowLang provides the necessary constructs and mechanisms for specifying such knowledge models at two main levels -- an ontology level and a logic-foundations level, where the latter is formed by special facts, rules, constraints and inter-ontology operators. In this paper, we also survey one of the ASCENS case studies to derive some of the requirements for KnowLang.

Supervisory control for software runtime exception avoidance

The Supervisory Control Theory (SCT) introduced by Ramadge and Wonham offers a framework for the control of Discrete Event Systems. In this paper, we formalize some concepts about corrective software maintenance within this framework. More specifically, we consider SCT as a way to control software systems behaviors and avoid occurrences of runtime exceptions. This approach is attractive as algorithms for controllers synthesis offer a means to automate part of the corrective maintenance process. In this paper, we introduce problems related to removing observed software failures by control, as well as solutions.

Execution trace exploration and analysis using ontologies

Dynamic analysis is the analysis of the properties of a running program. In order to perform dynamic analysis, information about the running program is often collected through execution traces. Exploring and analyzing these traces can be an issue due to their size and that knowledge of a human expert is often needed to derive the required conclusions. In this paper we provide a framework in which the semantics of execution traces, as well as that of dynamic analyses, are formally represented through ontologies. In this framework the exploration and analysis of the traces is enabled through semantic queries, and enhanced further through automated reasoning on the ontologies. We will also provide ontologies to represent traces and some basic dynamic analysis techniques, along with semantic queries that enable these techniques. Finally we will illustrate our approach through an example.

An approach for modeling dynamic analysis using ontologies

In this paper we present the possibility of using an ontology based framework in order to model Dynamic Analysis techniques. This work relies on similar ideas applied to the case of Static Analysis [22, 28, 27], in which ontologies are used to represent some knowledge about the programs to be analyzed. In the approach proposed in this paper we describe how ontologies can be applied to Dynamic Analysis by modeling both the information collected from the system, as well as some requirements about the type of analysis to be performed. Both of these ontologies can be designed by integrating ontologies previously defined during the software development cycle, allowing for re-usability. Finally, these ontologies make it possible to reason about concepts related to Dynamic Analysis and offer tools that facilitate automation. This paper presents the main ideas of the proposed approach and illustrates them with an example related to Frequency Spectrum Analysis.