Lukas Esterle

ARES: Adaptive Receding-Horizon Synthesis of Optimal Plans

We introduce ARES, an efficient approximation algorithm for generating optimal plans action sequences that take an initial state of a Markov Decision Process MDP to a state whose cost is below a specified convergence threshold. ARES uses Particle Swarm Optimization, with adaptive sizing for both the receding horizon and the particle swarm. Inspired by Importance Splitting, the length of the horizon and the number of particles are chosen such that at least one particle reaches a next-level state, that is, a state where the cost decreases by a required delta from the previous-level state. The level relation on states and the plans constructed by ARES implicitly define a Lyapunov function and an optimal policy, respectively, both of which could be explicitly generated by applying ARES to all states of the MDP, up to some topological equivalence relation. We also assess the effectiveness of ARES by statistically evaluating its rate of success in generating optimal plans. The ARES algorithm resulted from our desire to clarify if flying in V-formation is a flocking policy that optimizes energy conservation, clear view, and velocity alignment. That is, we were interested to see if one could find optimal plans that bring a flock from an arbitrary initial state to a state exhibiting a single connected V-formation. For flocks with 7 birds, ARES is able to generate a plan that leads to a V-formation in 95% of the 8,000 random initial configurations within 63i¾?s, on average. ARES can also be easily customized into a model-predictive controller MPC with an adaptive receding horizon and statistical guarantees of convergence. To the best of our knowledge, our adaptive-sizing approach is the first to provide convergence guarantees in receding-horizon techniques.

The Notion of Self-aware Computing

We define the notion of “self-aware computing” and the relationship of this term to related terms such as autonomic computing, self-management, and similar. The need for a new definition, driven by trends that are only partially addressed by existing areas of research, is motivated. The semantics of the provided definition are discussed in detail examining the selected wording and explaining its meaning to avoid misleading interpretations. This chapter also provides an overview of the existing usage of the term self-aware computing, respectively self-awareness, in related past projects and initiatives.

Self-adaptation for Individual Self-aware Computing Systems

This chapter discusses the role of self-awareness for adaptation at the individual level, when one single entity receives inputs both from itself or some of its components and from the external environment and uses the input to adjust to the current conditions. The chapter reviews the most widely used techniques for self-adaptation and identifies the role of self-awareness for each of the techniques and the metrics used to evaluate these techniques. Finally, we pave the way toward the following chapter, which discusses multiple entity adaptation, by introducing the interaction of different self-adaptation techniques at the level of the single individual.

Attacking the V: On the Resiliency of Adaptive-Horizon MPC

Inspired by the emerging problem of CPS security, we introduce the concept of controller-attacker games. A controller-attacker game is a two-player stochastic game, where the two players, a controller and an attacker, have antagonistic objectives. A controller-attacker game is formulated in terms of a Markov Decision Process (MDP), with the controller and the attacker jointly determining the MDP’s transition probabilities. We also introduce the class of controller-attacker games we call V-formation games, where the goal of the controller is to maneuver the plant (a simple model of flocking dynamics) into a V-formation, and the goal of the attacker is to prevent the controller from doing so. Controllers in V-formation games utilize a new formulation of model-predictive control we have developed called Adaptive-Horizon MPC (AMPC), giving them extraordinary power: we prove that under certain controllability conditions, an AMPC controller can attain V-formation with probability 1. We evaluate AMPC’s performance on V-formation games using statistical model checking. Our experiments demonstrate that (a) as we increase the power of the attacker, the AMPC controller adapts by suitably increasing its horizon, and thus demonstrates resiliency to a variety of attacks; and (b) an intelligent attacker can significantly outperform its naive counterpart.

Self-adaptation in Collective Self-aware Computing Systems

The goals of this chapter are to identify the challenges involved in self-adaptation (including learning and knowledge sharing) of multiple self-aware systems (or system collectives). We shall discuss the techniques available for dealing with the challenges identified (e.g., algorithms for conflict resolution, collective learning, and negotiation protocols), and which are appropriate given assumptions regarding the collective system architecture. We refer to notions of knowledge, learning, and adaptation; various self-awareness levels; and reference scenarios introduced in Chap. 4.

Towards a Framework for the Levels and Aspects of Self-aware Computing Systems

Increased self-awareness in computing systems can be beneficial in several respects, including a greater capacity to adapt, to build potential for future adaptation in unknown environments, and to explain their behaviour to humans and other systems. When attempting to endow computing systems with a form of self-awareness, it is important to have a clear understanding of what that form looks like. This chapter therefore first introduces the general concept of self-awareness and its various facets. Second, we provide an overview of the range of efforts to interpret the concept of self-awareness in computing. Third, we provide a structured conceptual framework that organizes this variety of different forms of self-awareness. This provides a broad set of concepts and a language that can be used to describe and reason about self-aware computing systems.

Transition Strategies for Increasing Self-awareness in Existing Types of Computing Systems

In this chapter, we propose a methodology to analyse the different levels of self-awareness present in distinct types of computing systems and architectures, investigate the level of self-awareness that is already present in those systems and architectures, and describe some transition strategies to increase the level of self-awareness in these systems.

Self-aware Computing Systems: Open Challenges and Future Research Directions

In this chapter, we discuss the open challenges in building self-aware computing systems that are still being faced by the research and development community. The challenges can be theoretical, technical, computational, or even sociological. First, we highlight the challenges associated with each of the earlier parts of the book and summarize on respective future research directions. We then offer concluding remarks and an outlook into the future in the last section.

Assessing Self-awareness

This chapter discusses the importance of assessing self-awareness of a system and different approaches and aspects on how to enable a human as well as a machine to perform such an assessment. The chapter also elaborates on the different requirements and constraints for an assessment. Furthermore, this chapter outlines how these requirements and constraints limit the accuracy of defining the capabilities of an assessed system and the corresponding degree of self-awareness.

Architectures for Collective Self-aware Computing Systems

This chapter aims to discuss the architectural aspects relevant to collectives of self-aware computing systems. Here, collectives consist of several self-aware computing systems that interact in some way. Their interactions may, potentially, lead to the formation of a self-aware collective of systems. Hence, the chapter defines different types of interactions that can link systems into a collective and then discusses the conditions under which self-awareness can be achieved within such collectives. Furthermore, the chapter identifies some of the most relevant architectural concerns that occur when linking multiple self-aware systems into a (self-aware) collective and defines these in the form of a generic meta-architecture for collectives of self-aware systems. Architectural concerns can represent both static and dynamic aspects of system collectives. Static concerns include the self-awareness levels of systems in a collective; the system interrelations, such as competition and cooperation; and several organisation patterns for systems in a collective, such as hierarchy or peer-to-peer designs. Dynamic concerns address changes that may occur over time, with respect to the above-mentioned aspects, based on the experience and learning of systems within the collective. More advanced topics discuss the manner in which the creation of collectives from interrelated systems can be applied recursively, adopting different architectural choices and combinations at each level, and potentially leading to a wide range of variations in the resulting self-awareness characteristics. The chapter concludes by indicating the main contributions and targeted beneficiaries of this chapter and points to the most important challenges to address in future research.