David B. Knoester

Applying genetic algorithms to decision making in autonomic computing systems

Increasingly, applications need to be able to self-reconfigure in response to changing requirements and environmental conditions. Autonomic computing has been proposed as a means for automating software maintenance tasks. As the complexity of adaptive and autonomic systems grows, designing and managing the set of reconfiguration rules becomes increasingly challenging and may produce inconsistencies. This paper proposes an approach to leverage genetic algorithms in the decision-making process of an autonomic system. This approach enables a system to dynamically evolve reconfiguration plans at run time in response to changing requirements and environmental conditions. A key feature of this approach is incorporating system and environmental monitoring information into the genetic algorithm such that specific changes in the environment automatically drive the evolutionary process towards new viable solutions. We have applied this genetic-algorithm based approach to the dynamic reconfiguration of a collection of remote data mirrors, with the goal of minimizing costs while maximizing data reliability and network performance, even in the presence of link failures.

Predator confusion is sufficient to evolve swarming behaviour

Swarming behaviours in animals have been extensively studied owing to their implications for the evolution of cooperation, social cognition and predator–prey dynamics. An important goal of these studies is discerning which evolutionary pressures favour the formation of swarms. One hypothesis is that swarms arise because the presence of multiple moving prey in swarms causes confusion for attacking predators, but it remains unclear how important this selective force is. Using an evolutionary model of a predator–prey system, we show that predator confusion provides a sufficient selection pressure to evolve swarming behaviour in prey. Furthermore, we demonstrate that the evolutionary effect of predator confusion on prey could in turn exert pressure on the structure of the predators visual field, favouring the frontally oriented, high-resolution visual systems commonly observed in predators that feed on swarming animals. Finally, we provide evidence that when prey evolve swarming in response to predator confusion, there is a change in the shape of the functional response curve describing the predators consumption rate as prey density increases. Thus, we show that a relatively simple perceptual constraint—predator confusion—could have pervasive evolutionary effects on prey behaviour, predator sensory mechanisms and the ecological interactions between predators and prey.

Harnessing Digital Evolution

In digital evolution, self-replicating computer programs-digital organisms-experience mutations and selective pressures, potentially producing computational systems that, like natural organisms, adapt to their environment and protect themselves from threats. Such organisms can help guide the design of computer software.

Automatically exploring how uncertainty impacts behavior of dynamically adaptive systems

A dynamically adaptive system (DAS) monitors itself and its execution environment to evaluate requirements satisfaction at run time. Unanticipated environmental conditions may produce sensory inputs that alter the self-assessment capabilities of a DAS in unpredictable and undesirable ways. Moreover, it is impossible for a human to know or enumerate all possible combinations of system and environmental conditions that a DAS may encounter throughout its lifetime. This paper introduces Loki, an approach for automatically discovering combinations of environmental conditions that produce requirements violations and latent behaviors in a DAS. By anticipating adverse environmental conditions that might arise at run time, Loki facilitates the identification of goals with inadequate obstacle mitigations or insufficient constraints to prevent such unwanted behaviors. We apply Loki to an autonomous vehicle system and describe several undesirable behaviors discovered.

Plato: a genetic algorithm approach to run-time reconfiguration in autonomic computing systems

Increasingly, applications need to be able to self-reconfigure in response to changing requirements and environmental conditions. Autonomic computing has been proposed as a means for automating software maintenance tasks. As the complexity of adaptive and autonomic systems grows, designing and managing the set of reconfiguration rules becomes increasingly challenging and may produce inconsistencies. This paper proposes an approach to leverage genetic algorithms in the decision-making process of an autonomic system. This approach enables a system to dynamically evolve target reconfigurations at run time that balance tradeoffs between functional and non-functional requirements in response to changing requirements and environmental conditions. A key feature of this approach is incorporating system and environmental monitoring information into the genetic algorithm such that specific changes in the environment automatically drive the evolutionary process towards new viable solutions. We have applied this genetic-algorithm based approach to the dynamic reconfiguration of a collection of remote data mirrors, demonstrating an effective decision-making method for diffusing data and minimizing operational costs while maximizing data reliability and network performance, even in the presence of link failures.

Digital Evolution of Behavioral Models for Autonomic Systems

We describe an automated method to generating models of an autonomic system. Specifically, we generate UML state diagrams for a set of interacting objects, including the extension of existing state diagrams to support new behavior. The approach is based on digital evolution, a form of evolutionary computation that enables a designer to explore an enormous solution space for complex problems. In our application of this technology, an evolving population of digital organisms is subjected to natural selection, where organisms are rewarded for generating state diagrams that support key scenarios and satisfy critical properties as specified by the developer. To achieve this capability, we extended the Avida digital evolution platform to enable state diagram generation, and integrated AviDA with third-party software engineering tools, e.g., the Spin model checker, to assess the generated state diagrams. To illustrate this approach, we successfully applied it to the generation of state diagrams describing the autonomous navigation behavior of a humanoid robot.

The evolutionary origin of somatic cells under the dirty work hypothesis.

Experimental evolution of digital organisms suggests that mutagenic side effects associated with performing valuable metabolic work can produce germ-soma differentiation in multicellular organisms.

Cooperative network construction using digital germlines

This paper describes a study in the evolution of cooperative behavior, specifically the construction of communication networks, through digital evolution and multilevel selection. In digital evolution, a population of self-replicating computer programs exists in a user-defined computational environment and is subject to instruction-level mutations and natural selection. Multilevel selection links the survival of the individual to the survival of its group, thus encouraging cooperation. The results of experiments using the Avida digital evolution platform demonstrate that populations of digital organisms are capable of constructing communication networks, and that these networks can exhibit desired properties depending on the selective pressures used. We also show that the use of a digital germline can significantly improve evolvability of cooperation.

Using group selection to evolve leadership in populations of self-replicating digital organisms

This paper describes a study in the evolution of distributed cooperative behavior, specifically leader election, through digital evolution and group selection. In digital evolution, a population of self-replicating computer programs exists in a user-defined computational environment and is subject to instruction-level mutations and natural selection. Group selection is the theory that the survival of the individual is linked to the survival of the group, thus encouraging cooperation. The results of experiments using the Avida digital evolution platform demonstrate that group selection can produce populations capable of electing a leader and, when that leader is terminated, electing a new leader. This result serves as an existence proof that group selection and digital evolution can produce complex cooperative behaviors, and therefore have promise in the design of robust distributed computing systems.

Evolution of swarming behavior is shaped by how predators attack

Animal grouping behaviors have been widely studied due to their implications for understanding social intelligence, collective cognition, and potential applications in engineering, artificial intelligence, and robotics. An important biological aspect of these studies is discerning which selection pressures favor the evolution of grouping behavior. In the past decade, researchers have begun using evolutionary computation to study the evolutionary effects of these selection pressures in predator-prey models. The selfish herd hypothesis states that concentrated groups arise because prey selfishly attempt to place their conspecifics between themselves and the predator, thus causing an endless cycle of movement toward the center of the group. Using an evolutionary model of a predator-prey system, we show that how predators attack is critical to the evolution of the selfish herd. Following this discovery, we show that density-dependent predation provides an abstraction of Hamiltons original formulation of domains of danger. Finally, we verify that density-dependent predation provides a sufficient selective advantage for prey to evolve the selfish herd in response to predation by coevolving predators. Thus, our work corroborates Hamiltons selfish herd hypothesis in a digital evolutionary model, refines the assumptions of the selfish herd hypothesis, and generalizes the domain of danger concept to density-dependent predation.