Intelligent Distributed Computing
Salvatore Venticinque, David Camacho · Concurrency and Computation Practice and Experience · 2016
Billions of computing elements continuously collect data and elaborate information nowadays. They range from personal computers to high-performance machines, from virtual machines to physical clusters and from smartphones to embedded systems. Most of them are connected, and this number increases continuously. Such an unlimited amount of spread resources offers challenging opportunities for building new kinds of computing overlay, for inferring collective knowledge and for developing emergent applications. In this context, the emergent field of Intelligent Distributed Computing focuses on the development of a new generation of intelligent distributed systems. It faces the challenges of adapting and combining research in the fields of Intelligent Computing and Distributed Computing. Intelligent Computing develops methods and technology ranging from classical artificial intelligence, computational intelligence and multi-agent systems to game theory. The field of Distributed Computing develops methods and technology to build systems that are composed of interacting and collaborating components. The International Symposium on Intelligent Distributed Computing (IDC) has a special interest in (but will not be limited to) novel architectures, systems and methods that facilitate distributed/parallel/multi-agent biocomputing for solving complex computational and real-life problems. The eighth edition of this conference was held in 2014 in Madrid under the auspicious of Autonomous University of Madrid (www.uam.es). A careful selection from some of the best paper presented at IDC2014 welcomes focused on ‘Intelligent Distributed Computing’ that were selected and invited to be extended for its potential publication at Concurrency and Computation: Practice and Experience journal. The special issue received submissions of original papers on all aspects of IDC ranging from concepts and theoretical developments to advanced technologies and innovative applications; some of the most relevant areas cover by this special issue includes the following: Intelligent Distributed High-performance Architectures; Organization and Management of Intelligent Distributed Systems; Intelligent Distributed Knowledge Representation and Processing; Networked and Distributed Intelligence and Intelligent Distributed Applications and Case Studies. From the received papers, those with the highest quality were selected and finally accepted. In the next section, a short description of final accepted papers is briefly outlined. Pokahr 8 focuses on the effective utilization of theoretically unlimited distributed and virtual resources, which are provided by the Cloud computing paradigm. In this context, intelligence is necessary to allow for exploiting the elastic computing infrastructure by a distributed application. This paper by Alexander Pokahr and Lars Braubach presents a component-based Cloud platform that provides autonomic management of applications using scale-out and on-demand deployment of computing resources at IaaS (Infrastructure as a Service) level. An implementation has been developed extending a multi-agent middleware towards a PaaS (Platform As A Service) Cloud infrastructure. Improving performance in distributed system is the objective of the paper 6. In fact, there are relevant issues also in data bound applications, because of the increasing number of users who are always connected and continuously publish contents in social networks or in their own remote repositories. Paper 6, authored by Randi Karlsen, David Sundby and Joan Nordbotten, deals with design and development of image retrieval algorithms of full set of thematic images, from huge collections from millions of users, which can be found in many social networks nowadays. The proposed solution can automatically generate an image collection description suitable for distributed search. This approach enhances the image tag sets collected by the host system for development of a collection description that provides an extended vocabulary to match search query terms. The retrieval process first selects collections that are relevant to the query, before retrieval of relevant images from those collections. This two-step process improves query processing efficiency, because irrelevant collections need not be searched. Annotation of contents with metadata is as relevant as difficult to achieve, because of a large amount of data exists on social networks services without annotations. An example of meta-information is geographical location of contents. It has become common for users to geotag resources on many online social networking services, but automatic annotation is still an open issue. Tri Nguyen Tuong, Dosam Hwang and Jason J. Jung 7 propose a method to predict the location of unlabeled resources on social networking services. The described approach uses the Naive Bayes and Support Vector Machine methods to classify the resources that are collected by using the term frequency of the tags in each class. Calculation for these methods is improved by using the values of the term frequency, and the class frequency is inverted to optimize the input data. These results can be applied to tag unlabelled resources on social networking services. Distributed intelligence can be exploited also on improved human users' performance in workplace contexts. In fact, usually, workers use computers connected to the network to carry out their tasks. This provides the possibility to collect information and plan actions to improve the development of working activities. In 3, Davide Carneiro, André Pimenta, Sérgio Gonçalves, José Neves and Paulo Novais discuss about monitoring and management of individual's performance in workplace contexts. The proposed approach is based on the observation of the worker's interaction with computer. Musical selection is investigated as an effective method for improving performance in the workplace. The described infrastructure allows team coordinators to assess and manage their co-workers' performance continuously and in real time, using a distributed service-based architecture. Performance results are discussed by experimental activities in real scenarios. We can understand, just from these works the relevance of a monitoring infrastructure. It is needed to build the necessary knowledge base for reasoning about effective actions to take for reconfiguration or optimization of distributed systems. Paper 5, authored by Kai Jander, Lars Braubach and Winfried Lamersdorf, focuses on distributed monitoring and workflow analysis and re-engineering of business processes. The research contribution deals with the utilization goal-oriented processes for modelling and executing workflow of collaborative business processes. The proposed solution allows for real-time association of occurring actions with business goals in the process. It uses drill-down analysis of events resulting from the execution of goal-oriented workflows in a distributed workflow environment After monitoring, learning becomes the next relevant problem. In fact, the availability of heterogeneous data, which have been collected by computers, smartphones and embedded devices, represent a precious source of information, which can be exploited in many application domains. Ricardo Aler, Ricardo Martin, Jose Valls and Ines Galván in 1 develop machine learning algorithms for forecasting of solar energy in the contexts of renewable energy sources. The prediction of solar energy is derived from numerical weather prediction models, which predicts meteorological variables for nodes in a grid. The paper investigates how prediction accuracy improves depending on the number of grid nodes and on the number and types of attributes. Many feature selection attributes are tested, and experimental results on historical data are used to build models for prediction of solar energy production in new locations. Distributed intelligence and exploitation of collective knowledge is the topic of the next research work. Authors use software agents to exploit collective intelligence for optimizing green energy utilization in smart-grids. In particular, the paper 2 authored by Alba Amato, Beniamino Di Martino, Marco Scialdone and Salvatore Venticinque focuses on a P2P network of users and agents for energy monitoring and management in smart solar powered micro-grids. Software agents are delegated to learn and predict energy profiles and to optimize the share of green energy within a neighbourhood. The emergent behaviour of the multi-agent system is an optimal schedule of consuming devices according to users' preferences and constraints. The platform design and the technological stack of the prototypal implementation are presented. The last cited papers demonstrate the proliferation of many kinds of social networks, where services are used by users, robots and even physical objects, which join the so-called Internet of Things. In such a cyber-physical system, security plays a key role. In the paper 9, Esther Villar-Rodriguez, Javier Del Ser and Sancho Salcedo-Sanz describe how this trend has given rise to a myriad of fraudulent strategies aimed at getting some sorts of benefit from the attacked individual. Stealing the credentials of the victim and assuming his or her identity to obtain access to resources (e.g. relationships or confidential information), credit and other benefits in that person's name are the objective of attackers. The paper delves into a machine learning approach that permits to efficiently detect this kind of attacks by solely relying on connection time information of the potential victim. Authors demonstrate how these learning algorithms – in particular, support vector classifiers – can be of great help to understand and detect impersonation attacks without compromising the user privacy of social networks. Because of problems like the ones presented previously, many mechanisms have been designed and implemented to increase the strength of the authentication mechanisms. A well-known example is Secure-images as text CAPTCHA. In this context paper 4 authored by Carlos Javier Hernández-Castro, David F. Barrero, María D. R-Moreno deals with the usage of Human Interactive Proofs (HIPs) to avoid automatic attacks. Civil Rights CAPTCHA is proposed here to aim at higher security. Empathy capacity of humans is exploited to further strengthen the security of text CAPTCHA. Fundamental design flaws are analysed from a security perspective using several well-known Machine Learning algorithms. Authors show that there is no need to solve the problem of neither OCR nor empathy analysis for computers to break this HIP. On the opposite, Machine Learning has been successfully used to break an HIP that uses both with a side-channel attack. This special issue has been achieved by a number of fruitful collaborations. We would like to thank the Editor in Chief of Concurrency and Computation: Practice and Experience, Prof. Geoffrey C. Fox, for his kind support and help during the whole process of publication. The guest editors would like to thank the reviewers for their valuable contribution, which has given this special issue the high quality we were expecting. Finally, this work has been partially supported by several research projects: Comunidad Autónoma de Madrid under project CIBERDINE S2013/ICE-3095 and by Spanish Ministry of Science and Education under project code TIN2014-56494-C4-4-P.