Antonella Di Stefano

A P2P strategy for QoS discovery and SLA negotiation in Grid environment

In recent years, Grid systems and Peer to Peer networks are the most commonly-used solutions to achieve the same goal: the sharing of resources and services in heterogeneous, dynamic, distributed environments. Many studies have proposed hybrid approaches that try to conjugate the advantages of the two models. This paper proposes an architecture that integrates the P2P interaction model in Grid environments, so as to build an open cooperative model wherein Grid entities are composed in a decentralized way. In particular, this paper focuses on a QoS aware discovery algorithm for P2P Grid systems, analyzing protocol and explaining techniques used to improve its performance.

An economic model for resource management in a Grid-based content distribution network

This paper presents an architecture to federate Content Distribution Networks (CDNs) in order to share computational resources, thus building an infrastructure that we call a Content Distribution Grid (CDG). The purpose of a CDG is to use a community policy allowing each CDN to put together a portion of its own resources to meet each others requirements, and therefore to guarantee a stronger quality of service to users. Since CDNs of a CDG belong to different organisations, the interaction scenario can be considered competitive, that is, organizations are mainly self-interested and are aiming at maximizing the performances of its own system. For this reason, the resource sharing policy proposed here is based on a offer/demand competitive model, in which resources are purchased by paying for them a certain amount of (virtual or real) money. An economic model is thus derived to guide such a sale; here the CDN requesting resources (buyer) and the CDN offering resources (seller) agree on the quantity and the price by means of a utility-based negotiation approach. A multi-agent system is then proposed to realise the software architecture supporting this model of CDG.

Supporting Agent Development in Erlang through the eXAT Platform

This work describes a new approach for writing multi-agent systems considering the use of the Erlang programming language. An analysis of the features of this language is provided, which shows that Erlang characteristics allow a programmer to easily model and implement agent systems. Then, a new agent programming platform, called eXAT—erlang eXperimental Agent Tool—will be described. This platform has been designed by the authors to support agent development and deploying with Erlang; the aim is to provide an all-in-one environment, allowing an agent designer to program agent intelligence, agent behavior and agent communication with a single language.

ERESYE: artificial intelligence in Erlang programs

This paper describes ERESYE, a tool for the realization of intelligent systems expert systems) using the Erlang language. ERESYE is a rule production system that allows rules to be written as Erlang function clauses, providing support for their execution. ERESYE is also able to support object-oriented concepts and ontologies thanks to a suitable ontology handling tool, providing means to translate object-based concepts into an Erlang form. The architecture of ERESYE and its basic working scheme are described in the paper. A comparison with CLIPS, one of the most known tools for expert system programming, is also made. The description of some examples of ERESYE usage are provided to show the effectiveness and the validity of the proposed solution, which opens new and interesting application scenario for Erlang.

An ACO Inspired Strategy to Improve Jobs Scheduling in a Grid Environment

Scheduling is one of the most crucial issue in a grid environment because it strongly affects the performance of the whole system. In literature there are several algorithms that try to obtain the best performance possible for the specified requirements; taking into account that the issue of allocating jobs on resources is a combinatorial optimization problem, NP-hard in most cases, several heuristics have been proposed to provide good performance. In this work an algorithm inspired to Ant Colony Optimization theory is proposed: this algorithm, named Aliened Ant Algorithm, is based on a different interpretation of pheromone trails. The goodness of the proposed algorithm, in term of load balancing and average queue waiting time, has been evaluated by mean of a vast campaign of simulations carried out on some real scenarios of a grid infrastructure.

A peer‐to‐peer decentralized strategy for resource management in computational Grids

This paper presents a peer‐to‐peer (P2P) approach for the management, in a computational Grid, of those resources that are featured by numerical quantity and thus characterized by a coefficient of utilization, such as percentage of CPU time, disk space, memory space, etc. The proposed approach exploits spatial computing concepts and models a Grid by means of a flat P2P architecture consisting of nodes connected by an overlay network; such a network topology, together with the quantity of resource available in each node, forms a three‐dimensional surface, where valleys correspond to nodes with a large quantity of available resource. In this scenario, this paper proposes an algorithm for resource discovery that is based on navigating such a surface, in search of the deepest valley (global minimum, that is, the best node). The algorithm, which aims at fairly distributing among nodes the quantity of leased resource, is based on some heuristics that mimic the laws of kinematics. Experimental results show the effectiveness of the algorithm. Copyright

A bio-inspired distributed algorithm to improve scheduling performance of multi-broker grids

The scheduling in grids is known to be a NP-hard problem. The distributed deployment of nodes, their heterogeneity and their fluctuations in terms of workload and availability make the design of an effective scheduling algorithm a very complex issue. The scientific literature has proposed several heuristics able to tackle this kind of optimization problem using techniques and strategies inspired by nature. The algorithms belonging to ant colony optimization (ACO) paradigm represent an example of these techniques: each one of these algorithms uses strategies inspired by the self-organization ability of real ants for building effective grid schedulers. In this paper, the authors propose an on line, non-clairvoyant, distributed scheduling solution for multi-broker grid based on the alienated ant algorithm (AAA), a new ACO inspired technique exploiting a “non natural” behavior of ants and an inverse interpretation of pheromone trails. The paper introduces the proposed algorithm, explains the differences with other classical ACO approaches, and compares AAA with two different algorithms. The results of simulations show that the AAA guarantees good performance in terms of makespan, average queue waiting time and load balancing capability.

A Multi-agent Reflective Architecture for User Assistance and Its Application to E-commerce

Assisting an user working with an application can involve several tasks of a different nature; thus it can be a complex job which is better performed by several autonomous agents. Accordingly, in many scenarios, several small assistant agents, each dedicated to a single task, are employed to supply help and to enhance the same application.This paper proposes a software architecture that allows multiple assistants to serve the same application and interact with each other as necessary, while working autonomously from each other. This architecture interfaces assistants with an existing application by means of computational reflection. The latter mechanism allows meaningful user activities to be intercepted by assistants, and the outcomes of their activity to be supplied to the application. No assumptions need to be made about the application or the assistants; assistants can be changed, added and removed as necessary to adapt the application to unforeseen scenarios, conversely an assistant can be employed to support several applications. The usefulness and applicability of the proposed architecture is demonstrated by an e-commerce case study: we show how a suitable assistant set can integrate with and enhance a bare web browser, making it fit to support e-commerce activities.

The transparent implementation of agent communication contexts

Agent Communication Contexts (ACCs) are virtual environments where agents may live and interact. In this way, as in a human society, interactions may be subject to conventions and laws depending on the context where they occur. For this to be possible, an ACC should embed the communication laws relevant to the intended class of agent applications and enforce them, as interactions among agents take place. Although context is a communication aspect relevant for all the agents of an application, its modelling should be, in principle, an orthogonal concern with respect to the design of the activities of each agent. Consistently with this view, this work advocates the separate development of, respectively, agent behaviour, and the interaction aspects constituting the context. The latter is first abstractly specified as a set of communication laws, then automatically implemented by a tool that generates the necessary ACC management and checking code from the specification. The appropriate portions of this code should be activated whenever an interaction between agents takes place, so as to ensure that (i) the constraints specified by the laws are respected by the interaction, and (ii) the actions some of the laws require are carried out before the interaction actually occurs. Moreover, this work proposes an infrastructure whereby ACC code is triggered at runtime, whenever agents interact with each other. No source code modification or recompilation is required for this. All is seamlessly accomplished by means of computational reflection, which transparently changes the meaning of the communication primitives normally used by agent programmers. Copyright

Enforcing agent communication laws by means of a reflective framework

Agent Coordination Contexts (ACCs) have been proposed as virtual environments where agents live and interact. In this way, as in a human society, interactions may be subjected to conventions and laws depending on their context. This is obtained by a suitable ACC that embeds the communication laws relevant to a specific application and checks whether they are fulfilled as interactions take place.Context modelling, while representing a communication aspect relevant for all the agents of an application, is a crosscutting concern with respect to the design of the activities of each agent. In this paper, we propose an approach allowing a separate design and implementation of, respectively, behaviour and the interaction aspects constituting the context. Once the latter have been formalised in a specification consisting of communication laws, a tool generates the necessary management and checking code from the specification.Moreover, we automate the way laws are enforced on agent communication by suitably redirecting any interaction between agents, so as to ensure that the constraints specified by the laws are respected by the interaction, and the actions required by some laws are taken before it actually takes place. Redirection is accomplished by means of computational reflection, which transparently changes the meaning of the communication primitives normally used by agents programmers.