Abstract—This paper proposes a novel framework for developing adaptive and scalable network services. In the proposed framework, a network service is implemented as a group of autonomous agents that interact in the network environment. Agents in the proposed framework are autonomous and capable of simple behaviors (e.g., replication, migration, and death). In this paper, an evolutionary adaptation mechanism is designed using genetic algorithms (GAs) for agents to evolve their behaviors and improve their fitness values (e.g., response time to a service request) to the environment. The proposed framework is evaluated through simulations, and the simulation results demonstrate the ability of autonomous agents to adapt to the network environment. The proposed framework may be suitable for disseminating network services in dynamic and large-scale networks where a large number of data and services need to be replicated, moved, and deleted in a decentralized manner. Index Terms—Adaptive and scalable network services, autonomous agents, evolutionary computation, self-organization, swarm intelligence. I.

Some recent successful techniques to solve multiobjective optimisation problems are based on variants of evolutionary algorithms and use recombination and self-adaptation to evolve the population. We present an approach that incorporates a population of solutions into a hybrid metaheuristic with no recombination. The population is evolved using self-adaptation, a mutation operator and an information-sharing mechanism. Since the main component in our approach is a simulated annealing algorithm, the cooling schedule for the whole population becomes critical. A common cooling schedule for the whole population is determined based on an evolutionary process. Results are presented using a real-world multiobjective combinatorial optimisation problem, namely space allocation with two conflicting criteria. These results suggest that this approach is a suitable alternative not only for combinatorial multiobjective optimisation problems, but also for obtaining a population of locally optima solutions in singleobjective optimisation problems. 1

Opponent(s)

Abstract. This paper studies the fuzzification of the Pareto dominance relation and its application to the design of Evolutionary Multi-Objective Optimization algorithms. A generic ranking scheme is presented that assigns dominance degrees to any set of vectors in a scale-independent, nonsymmetric and set-dependent manner. Based on such a ranking scheme, the vector fitness values of a population can be replaced by the computed ranking values (representing the ”dominating strength ” of an individual against all other individuals in the population) and used to perform standard single-objective genetic operators. The corresponding extension of the Standard Genetic Algorithm, so-called Fuzzy-Dominance-Driven GA (FDD-GA), will be presented as well. To verify the usefulness of such an approach, an analytic study of the Pareto-Box problem is provided, showing the characteristical parameters of a random search for the Pareto front in a unit hypercube in arbitrary dimension. The basic problem here is the loss of dominated points with increasing problem dimension, which can be successfully resolved by basing the search procedure on the fuzzy dominance degrees. 1

Abstract — In this paper, we present an approach for improving fault-tolerance and service availability in intelligent video surveillance (IVS) systems. A typical IVS system consists of various intelligent video sensors that combine image sensing with video analysis and network streaming. System monitoring and fault diagnosis followed by appropriate dynamic system reconfiguration mitigate effects of faults and therefore enhance the system’s fault-tolerance. The applied monitoring and diagnosis unit (MDU) allows the detection of both node- and system-level faults. Lacking redundant hardware such reconfigurations are established by graceful degradation of the overall application. An optimizer module that performs multi-criterion optimization is used to compute a new degraded system configuration by trading off quality of service (QoS), energy consumption, and service availability. We demonstrate the functionality of our approach by an illustrative example. 1

Optimisation (MOO) has been focused on achieving the Pareto optimal front by explicitly analysing the dominance level of individual solutions. While such approaches have produced good results for a variety of problems, they are computationally expensive due to the complexities of deriving the dominance level for each solution against the entire population. TB_MOO (Threshold Based Multiobjective Optimisation) is a new artificial life approach to MOO problems that does not analyse dominance, nor perform any agent-agent comparisons. This reduction in complexity results in a significant decrease in processing overhead. Results show that TB_MOO performs comparably, and often better, than its more complicated counter-parts with respect to distance from the Pareto optimal front, but is slightly weaker in terms of distribution and extent. 1

Abstract. This paper presents an application of an evolutionary optimization algorithm for multiobjective analysis of selective withdrawal from a thermally stratified reservoir. A multiobjective particle swarm optimization (MOPSO) algorithm is used to find nondominated (Pareto) solutions when minimizing deviations from outflow water quality targets of: (i) temperature; (ii) dissolved oxygen (DO); (iii) total dissolved solids (TDS); and (iv) potential of hydrogen (pH). The decision variables are the flows through each port in the selective withdrawal structure. The MOPSO algorithm, implemented as an add-in for Excel, is able to find nondominated solutions for any combination of the four abovementioned objectives. An interactive graphical method was also developed to display nondominated solutions in such way that the best compromise solutions can be identified for different relative importance given to each objective. The method allows the decision maker to explore the Pareto set and visualize not only the best compromise solution but also sets of solutions that provide similar compromises. 1.

In recent years, the methods of evolutionary computation have proven themselves useful in the area of data mining. For rule mining, several objective functions have been used, relating to both accuracy and interestingness in general. However, when searching for rules or patterns in a data set, several conflicting objectives will often be present. As the ultimate goal of data mining is to discover unexpected, useful knowledge, it may not be feasible to prioritize these objectives a priori. Simply constructing an aggregate fitness function in these cases could be seen as a more or less ad hoc solution. In this paper we propose an alternative: Using well-established multiobjective evolutionary algorithms to evolve a Pareto optimal set of rules.

Abstract not found

Different types of niching can be used in genetic algorithms (GAs) or evolutionary computations (ECs) to sustain the diversity of the sought optimal solutions and to increase the effectiveness of evolutionary multi-objective optimization solvers. In this paper four schemes of niching are proposed, which are also considered in two versions with respect to the method of invoking: a continuous realization and a periodic one. The characteristics of these mechanisms are discussed, while as their performance and effectiveness are analyzed by considering exemplary multi-objective optimization tasks both of a synthetic and an engineering (FDI) design nature. Keywords: niching, ranking, Pareto-optimality, genetic algorithms, evolutionary computations, multi-objective optimization, solutions diversity, engineering designs, detection observers