skip to main content
10.1145/1809018acmconferencesBook PagePublication PagesicacConference Proceedingsconference-collections
BADS '10: Proceedings of the 2nd workshop on Bio-inspired algorithms for distributed systems
ACM2010 Proceeding
Publisher:
  • Association for Computing Machinery
  • New York
  • NY
  • United States
Conference:
ICAC '10: 7th International Conference on Autonomic Computing Washington DC USA 11 June 2010
ISBN:
978-1-4503-0086-5
Published:
11 June 2010
Sponsors:
In-Cooperation:
IEEE, University of Arizona
Recommend ACM DL
ALREADY A SUBSCRIBER?SIGN IN

Reflects downloads up to 13 Jan 2025Bibliometrics
Skip Abstract Section
Abstract

Evolution is an efficient and robust genetic operator that is incorporated into a Markov Chain Monte Carlo sampler to drive the evolution of the Markov Chain towards optimal solutions, in the presence of multiple objectives. The effectiveness of the proposed optimization technique is assessed in a set of test functions and is its superiority over other algorithms is demonstrated. The proposed method is shown to be easily adopted for execution on many-core architectures with GPU processors.

The following paper, by L. Vanneschi et al., presents four new parallel and distributed Particle Swarm Optimization (PSO) methods and two new sets of test functions with different difficulties, tuneable by modifying some parameters. The performance of the different PSO methods are compared on the two sets of test functions and on the well known set of Rastrigin test functions. The experimental results show that one of the four methods, namely the "multi-swarm repulsive PSO" (MRPSO), outperforms the others on the test problems. This is probably due to the higher diversity degree that the characteristics of MSRPO help to maintain in the system.

The session closes with the keynote talk, given by Kenneth De Jong, entitled "Evolutionary Computation: Challenges for the 21st Century". The field of Evolutionary Computation (EC) has experienced tremendous growth over the past 40 years, resulting in a wide variety of evolutionary algorithms and applications. This poses two interesting challenges: to provide a unified view of the field and to identify important open issues. The keynote talk addresses these challenges by giving an overview of a general EC framework that can help compare and contrast approaches, encourages crossbreeding, and facilitates intelligent design choices. The use of the framework is then illustrated by showing how traditional Evolutionary Algorithms can be compared and contrasted with it, and how new algorithms can be effectively designed. Finally, the framework is used to identify some important open issues that need further research.

The session on Bio-inspired Approaches for P2P and Cloud Computing comprises four technical papers. The paper by M. J. Csorba et al. presents a heuristic and decentralized optimization method aiming to find a suitable mapping between Virtual Machine (VMs) replicas and nodes in a hybrid private/public Cloud environment. The proposed heuristic is based on a variant of Ant Colony Optimization based on crossentropy. The algorithm uses a collection of ant agents that move around aiming to find efficient deployment mappings for a large number of virtual machine image replicas that are deployed concurrently, and at the same time to match important requirements like scalability and load balancing. The method is tested and assessed in a complex environment consisting of five interconnected Clouds.

The paper by C. Lee et al. presents a novel approach, named Nuage, to the important problem of mapping applications to hosts, with a particular reference to the Cloud environment. The approach is based on the theory of evolutionary games: tentative mapping schemes are associated to game strategies that are played by opponent players, members of an evolving population, to progressively construct efficient solutions. In the paper it is theoretically proved that the Nuage algorithm leads to stable equilibrium solutions, whereas simulation experiments show that the important adaptability property is also ensured, since resource allocation adapts to workload and resource availability.

The paper by R. N. Lass et al. presents a self-organizing framework for distributed computing in a heterogeneous scenario. The paper describes the design and implementation issues related to GUMP, an environment that helps developers to create network agnostic server-side proxies and support client-server sessions in dynamic P2P environments, with a specific focus on the requirements of wireless networks. The authors aim to make GUMP completely self-adaptive by using heuristic measurements to analyze/predict the mobility of the nodes in the network and to automatically choose the appropriate protocols and deployment stack that are the most appropriate to dynamically address the needs of the network.

The paper by C. Villalba and F. Zambonelli discusses how pervasive service systems, made up of a massive number of components, can get inspiration from natural systems, in particular ecological systems, by modeling and deploying services as autonomous agents, spatially situated in an ecosystem of other services, data sources, and pervasive devices. The authors introduce an approach that abstracts the components of the ecosystem as goal-oriented organisms that, driven by laws of survival, interact with each other and selforganize their activities according to dynamic food-web relations. The authors clarify the discussed concepts for a representative case study in the area of adaptive displays ecosystems, and assess the devised approach through a set of simulation experiments.

The content of workshop papers confirms the wide range of practical applications that may benefit from the adoption of bio-inspired algorithms, and also the inherent capacity of these algorithms to be exploited in emerging domains such as P2P networks and modern multi-core processors. The main reason is that bioinspired paradigms provide such characteristics as decentralization, self-organization, flexibility and energy saving that are essential to efficiently cope with the ever increasing complexity of current and future distributed systems. In particular, it is very interesting to notice that this community has quickly devised bioinspired techniques to cope with the issues raised by the recent development and usage of the Cloud paradigm. Cloud computing platforms, like Amazon Elastic Cloud Computing and Simple Storage Service, Google App Engine, and Microsoft Azure, supply computational and storage resources at an accessible cost, which were before available only for big companies and for large research groups. Several papers in this workshop confirm the benefits that bio-inspired algorithms can bring to Cloud Computing solutions. On the other hand, it remains to discover if the performance of bio-inspired applications can be further improved by new ways of designing them and by this new illusion of infinite computing brought by the use of Cloud Computing platforms.

Skip Table Of Content Section
SESSION: Evolutionary techniques for distributed systems
research-article
GPU-accelerated differential evolutionary Markov Chain Monte Carlo method for multi-objective optimization over continuous space

In this paper, the attractive features of evolutionary algorithm and Markov Chain Monte Carlo are combined into a new Differential Evolutionary Markov Chain Monte Carlo (DE-MCMC) method for multi-objective optimization problems with continuous ...

research-article
A study of parallel and distributed particle swarm optimization methods

The goal of this paper is to present four new parallel and distributed particle swarm optimization methods and to experimentally compare their performances on a wide set of benchmark functions. These methods include a genetic algorithm whose individuals ...

keynote
Evolutionary computation: challenges for the 21st century

The field of Evolutionary Computation has experienced tremendous growth over the past 40 years, resulting in a wide variety of evolutionary algorithms and applications. This poses two interesting challenges: to provide a unified view of the field and to ...

SESSION: Bio-inspired approaches for P2P and cloud computing
research-article
Ant system for service deployment in private and public clouds

Large-scale computing platforms that serve thousands or even millions of users through the Internet are on a path to become a pervasive technology available to companies of all sizes. However, existing technologies to enable this kind of scaling are ...

research-article
An evolutionary game theoretic approach to adaptive and stable application deployment in clouds

This paper studies an evolutionary game theoretic mechanism for adaptive and stable application deployment in cloud computing environments. The proposed mechanism, called Nuage, allows applications to adapt their locations and resource allocation to the ...

research-article
GUMP: adapting client/server messaging protocols into peer-to-peer serverless environments

In this paper we present a generic environment for creating message-oriented server-side proxies to support adaptation from TCP transport-oriented client-server sessions to many-to-many peer-to-peer networking environments more suitable for deployment ...

research-article
Simulation experiences with an ecological approach for pervasive service systems

Innovative frameworks have to be identified for the deployment and execution of pervasive service systems. These systems, composed by a massive number of components, should be able to exhibit properties of self-organization and self-adaptability, and of ...

Contributors
  • Italian National Research Council
  • Hellenic Mediterranean University
  • Istituto Di Calcolo E Reti Ad Alte Prestazioni, Rende
  • University of Massachusetts Boston

Recommendations