[pt] Sistemas multi-agentes são sociedades com agentes autônomos e heterogêneos que podem trabalhar em conjunto para alcançar objetivos similares ou totalmente diferentes. Quando falhas acontecem enquanto algum agente de software tenta alcançar seus objetivos, torna-se importante entender porque tais falhas acontecem e o que pode ser feito para remediar tais problemas. Considerando o ambiente distribuído, dinâmico e a natureza dos sistemas multi-agentes, é importante definir os requisitos necessários para realizar diagnósticos das falhas e recomendações de planos alternativos para agentes que desejam alcançar seus objetivos. Assim, esta dissertação propõe soluções para os principais desafios de criar um sistema que realize diagnósticos e proveja recomendações sobre execuções de agentes. Além disso, o trabalho propõe um framework híbrido de diagnóstico e recomendação que provê suporte para tais desafios. A partir do framework, instâncias de diferentes domínios podem ser criadas, como, por exemplo, aplicações baseadas em computação ubíqua e diferentes diagnósticos e recomendações podem ser providas. / [en] Multi-agent systems are societies with autonomous and heterogeneous agents that can work together to achieve similar or different goals. Agents executing in such systems may not be able to achieve their goals due to failures during system execution. When an agent tries to achieve its desired goals, but faces failures during execution, it becomes important to understand why such failures occurred and what can be done to remedy the problem. The distributed, dynamic and nature of multi-agent systems calls for a new form of failure handling approach to address its unique requirements, which involves both diagnosing specific failures and recommending alternative plans for successful agent execution and goal attainment. We discuss solutions to the main challenges of creating a system that can perform diagnoses and provide recommendations about agent executions to support goal attainment, and propose a hybrid diagnostic-recommendation framework that provides support for methods to address such challenges. From the framework, instances of different domains can be created, such as, applications based on ubiquitous computing and different diagnoses and recommendations can be provided.

[pt] A complexidade e a distribuição de sistemas multiagentes tornam seu desenvolvimento muito sujeito a falhas. Ao mesmo tempo, poucas iniciativas procuram resolver o problema através de teste automatizado da mesma maneira que no desenvolvimento orientado a objetos. Este trabalho apresenta um framework para o desenvolvimento de aplicações de teste automatizado através de agentes Stubs para o middleware M-Law. O middleware implementa um interpretador para a linguagem XMLaw, atuando como mediador da comunicação dos agentes segundo a definição das leis de interação. / [en] Complexity and distribution of multi-agent systems make the development of such systems failure prone. In this context, few solutions try to solve the problem through automated tests as we see in object oriented development. This work presents a framework for test automated applications development through the use of Stub agents for the middleware M-Law. This middleware implements the law-enforcement mechanism for regulating OMAS specified by the description language XMLaw which implements a law definition language, called XMLaw, for agent interaction in open multi-agent systems.

[pt] Os últimos anos têm sido marcados pelo significativo crescimento em todo o mundo da demanda por serviços de telecomunicações. Tal cenário de expansão de redes e da necessidade de coexistência e interoperabilidade de diferentes tecnologias de forma economicamente viável proporciona grandes desafios para a gerência, operação e manutenção de redes de telecomunicações. O presente trabalho apresenta alguns dos principais modelos e paradigmas de gerência de redes tradicionalmente empregados em redes de telecomunicações e que ainda hoje são amplamente utilizados pela indústria. Muitos dos modelos apresentados foram significativamente influenciados por conceitos e técnicas oriundos da área de engenharia de software. Uma grande ênfase é dada particularmente ao uso de técnicas baseadas em agentes de software para gerência de redes. Para tanto, importantes conceitos sobre agentes de software são apresentados, assim como exemplos de trabalhos em que agentes de software são utilizados no domínio de gerência de redes. Por fim, é proposta uma arquitetura baseada em agentes de software para gerência de falhas em redes legadas de telecomunicações, que são comumente gerenciadas por sistemas centralizados. O objetivo principal desta arquitetura é permitir o diagnóstico e a correção de falhas de rede de forma a não sobrecarregar o sistema centralizado de gerência. Para tanto, são utilizados agentes de software que distribuem informações mantidas no sistema centralizado para outros agentes do sistema. Desta forma, é possível que os agentes responsáveis por executar os procedimentos de diagnóstico e correção de falhas desempenhem suas atividades sem a necessidade de uma comunicação direta com o sistema centralizado. / [en] The last few years have been marked by a significant and worldwide growth in the demand for telecommunications services. Such scenery of network expansion and the need for coexistence and interoperability of different technologies in an economically viable way provides great challenges for the management, operation and maintenance of telecommunications networks. This work presents some of the main network management models and paradigms traditionally employed in telecommunications networks and that still count with wide adoption in the industry as of this day. Many of the presented models have been significantly influenced by concepts and techniques originated in the software engineering field. A great emphasis is particularly given to the use of network management techniques based on software agents. To this end, important concepts of software agents are presented, as well as examples of works where software agents are used in the network management domain. Finally, an architecture based on software agents used for fault management in legacy telecommunications networks, which are usually managed by centralized systems, is proposed. The main objective of this architecture is to allow the diagnosis and the correction of network faults in a way not to overload the centralized management system. To this end, the architecture uses software agents that distribute information maintained in the centralized management system to other agents of the system. In such way, it is possible for the agents responsible for executing the fault diagnosis and correction procedures to perform their activities without the necessity for direct communication with the centralized system.

Nous nous intéressons dans cette thèse à la planification pour les problèmes de prise de décision décentralisée séquentielle dans l'incertain. Dans le cadre centralisé, l'utilisation des formalismes MDP et POMDP a permis d'élaborer des techniques de planification efficaces. Le cadre Dec-POMDP permet de formaliser les problèmes décentralisés. Ce type de problèmes appartient à une autre classe de complexité que les problèmes centralisés. Pour cette raison, jusqu'à récemment, seuls de très petits problèmes pouvaient être résolus et uniquement pour des horizons très faibles. Des algorithmes heuristiques ont récemment été proposés pour traiter des problèmes de taille plus conséquente mais n'ont pas de preuve théorique de qualité de solution. Nous montrons comment une information heuristique sur le problème à résoudre représentée par une distribution de probabilité sur les croyances centralisées permet de guider la recherche approchée de politique. Cette information heuristique permet de formuler chaque étape de la planification comme un problème d'optimisation combinatoire. Cette formulation conduit à des politiques de meilleure qualité que les approches existantes. / In this thesis, we focus on planning in decentralised sequentialdecision taking in uncertainty. In the centralised case, the MDP andPOMDP frameworks leads to efficient planning algorithms. The Dec-POMDPframework is used to model decentralised problems. This kind ofproblems is in a higher class of complexity than the centralisedproblem. For this reason, until recently, only very small problem could be solved and only for very small horizons. Recently, some heuristic algorithms have been proposed to handle problem of higher size but there is no theoretic proof of the solution quality. In this thesis, we show how to use a heuristic information in the problem, modelled as a probability distribution on the centralised beliefs, to guide the search for a good approximate policy. Using this heuristic information, we formulate each time step of the planning procedure as a combinatorial optimisation problem. This formulation leads to policies of better quality than previously existing approaches.

[pt] Esta dissertação propõe um framework de sistemas multi- agentes para aplicações do domínio de Radio Frequency Identification (RFID). Esta tecnologia de identificação automática de objetos vem se desenvolvendo rapidamente e promete substituir brevemente a tecnologia atual de identificação de produtos: código de barras. Além de promover o desenvolvimento de hardware para etiquetas e leitoras RFID, novos padrões de identificação estão sendo propostos, desafiando os desenvolvedores de software a produzirem novas soluções para lidar com o novo e maior volume de dados e com a natureza distribuída das leitoras e leituras de objetos. Esta dissertação propõe um framework projetado utilizando-se a abordagem de sistemas multi-agentes (SMA) capaz de instanciar SMA distribuídos que: - Interagem com a infra-estrutura RFID existente (leitoras, etiquetas e middlewares); - Facilitam a geração de informações com semântica de negócios a partir de dados crus gerados por esta infra-estrutura, e; - Facilitam a implementação de novas funcionalidades decorrentes do leque de aplicações vislumbradas pelo surgimento de RFID. / [en] Radio Frequency Identification (RFID) technology along with the Electronic Product Code (EPC) created a new complex environment for warehouses systems and supply chain systems in general. The EPC specification created one more level of identification: the product instance. This new level of identification had blown the volume of data being managed by applications. That fact, along with the deployment of distributed RFID readers, causes the necessity of new applications capable of interacting with this infrastructure and capable of processing, filtering and interpreting these new data. This work proposes a framework for distributed multi-agent systems capable of interacting with a deployed RFID infrastructure.

Este trabalho apresenta o projeto e arquitetura de uma plataforma para execução de Agentes com funções implementadas em hardware, tomando vantagem do uso de hardware reconfigurável. Os Agentes em hardware são implementados utilizando dispositivos FPGA (Field-programmable Gate Array). O trabalho estende trabalhos anteriores semelhantes na área, com o diferencial de adicionar suporte às funcionalidades de reconfiguração parcial do hardware, suportar aplicações que demandam alto desempenho em hardware – como processamento de sinais e imagens – e redução de recursos de hardware necessários para execução da interface em software. A plataforma proposta utiliza o framework JADE (Java Agent Development Framework), que é um dos frameworks mais populares no estado da arte de desenvolvimento de Agentes e compatível com outros frameworks de Agentes através da conformidade aos padrões FIPA (Foundation for Intelligent Physical Agents). Com o uso do JADE, a plataforma possibilita a comunicação entre Agentes com funções implementadas em hardware e Agentes puramente implementados em software dentro de um mesmo SMA (Sistema Multi-Agente). Uma funcionalidade notável do JADE é a possibilidade de migração de Agentes entre plataformas de um mesmo SMA. Através do uso da reconfiguração parcial de hardware em conjunto com o JADE, a plataforma permite a migração de Agentes de software para hardware e vice-versa, além de suportar reconfiguração de múltiplos Agentes em hardware com um único FPGA. A plataforma faz uso de um único chip através do uso de um processador soft core implementado na lógica programável. O uso deste processador é um diferencial neste trabalho, e mostra que é possível utilizar o JADE em sistemas embarcados com recursos de processamento limitados. Ou seja, em um Agente cuja principal função é implementada em hardware, basta um processador bastante simples para atuar como uma interface entre o hardware e o framework de Agentes. O uso do processador dentro do FPGA tem também o benefício de oferecer formas de acesso mais integrado ao hardware, permitindo maior desempenho na transmissão de dados ao hardware. A plataforma foi validada através de estudos de caso de Agentes com implementações em hardware e em software, incluindo um estudo de caso aplicado de processamento de imagem embarcado utilizando VANTs (Veículos Aéreos Não-Tripulados). O estudo também apresenta comparações de desempenho entre a execução dos Agentes em hardware e em outras plataformas embarcadas de prateleira. Os experimentos realizados mostram um ganho significativo de desempenho nas implementações em FPGA, especialmente considerando processamento de imagens de alta resolução, mesmo considerando que o FPGA executa em frequências consideravelmente reduzidas em comparação às outras plataformas testadas. / This work described the design and architecture of a platform for execution of Agents whose functions are implemented in hardware, by leveraging the use of reconfigurable hardware. The hardware Agents are implemented using FPGA (Field-programmable Gate Array) devices. This work extends previous similar work in this field, while adding the features of hardware partial reconfiguration, supporting applications which require high performance in hardware – such as image or signal processing – and reducing the hardware resource for the software interface execution. The proposed platform makes use of the JADE (Java Agent Development Framework) framework, which is one of the most popular frameworks in state-of-the-art Agent development, and is also compatible with other Agent development frameworks due to compliance with FIPA (Foundation for Intelligent Physical Agents) standards. With the use of JADE, the platform enables communication among Agents which are implemented in hardware and Agents purely implemented in software, inside the same MAS (Multi-Agent System). One notable feature of JADE is the possibility of migrating Agents among platforms inside a single MAS. Through the use of hardware partial reconfiguration along with JADE, the platform enables the migration of Agents from software to hardware and viceversa, in addition to supporting múltiple hardware Agents in a single FPGA. The platform makes use of a single chip, by using a MicroBlaze soft core processor implemented in programmable logic. The use of this processor is a distinction on this work, and it shows that it is possible to use JADE on embedded systems with limited processing power. That is, in an Agent whose main function is implemented in hardware, a very simple processor to act as an interface between hardware and the Agent framework is enough. The use of the soft core processor inside the FPGA also has the benefit of offering more integrated ways of accessing hardware, enabling higher performance for transferring data to hardware. The platform was validated through case studies of hardware and software Agent implementation, including a case study applied to image processing using UAVs (Unmanned Aerial Vehicles). The study also shows performance comparisons between the Agent execution in hardware and in other off-the-shelf embedded platforms. The performed experiments report a significative performance increase in the FPGA implementations, particularly in high resolution image processing, even considering that the FPGA runs in considerably lower clock frequency than the other tested platforms.

Network Functions Virtualisation (NFV) separa as funções de rede dos dispositivos físicos, simplificando a implantação de novos serviços. As típicas funções de rede, como firewalls, aceleradores de tráfego, sistemas de detecção de intrusão e sistemas de prevenção de intrusões, são tradicionalmente realizadas por equipamentos físicos proprietários, que devem ser instalados manualmente pelos operadores de rede. A implantação de equipamentos físicos é desafiadora porque eles têm requisitos específicos de encadeamento e ordenação. Ao contrário dos equipamentos físicos tradicionais, as funções de rede virtuais (VNFs) podem ser dinamicamente implementadas e reconfiguradas sob demanda, colocando desafios de gerenciamento rigorosos aos sistemas em rede. A seleção das VNFs mais apropriadas para atingir um objetivo específico e a decisão sobre onde implantar essas VNFs e por quais caminhos elas se comunicarão são responsabilidades de um orquestrador de NFV. Nesta dissertação, propomos orquestrar VNFs usando componentes cognitivos interativos estruturados com a arquitetura belief-desire-intention (BDI), levando a soluções emergentes para enfrentar os desafios da rede. A arquitetura BDI inclui um ciclo de raciocínio que fornece aos agentes um comportamento racional, permitindo que lidem com diferentes cenários nos quais o comportamento flexível e inteligente é necessário. Estendemos a arquitetura NFV substituindo seu orquestrador centralizado por agentes BDI. Nossa proposta inclui um protocolo de leilão reverso e uma nova heurística de licitação que permite que os agentes tomem decisões sobre as tarefas de orquestração. Por fim, nós fornecemos uma plataforma de testes que integra uma plataforma para o desenvolvimento de agentes BDI com um emulador de rede, permitindo que os agentes controlem as VNFs e percebam a rede. Essa plataforma de testes é usada para implementar VNFs e avaliar empiricamente nosso modelo teórico em um ataque de negação de serviço distribuído. Os resultados da avaliação mostram que uma solução para o ataque DDoS surge através da negociação de agentes, mitigando com sucesso o ataque. / Network Functions Virtualisation (NFV) decouples network functions from physical devices, simplifying the deployment of new services. Typical network functions, like firewalls, traffic accelerators, intrusion detection systems and intrusion prevention systems, are traditionally performed by proprietary physical appliances, which must be manually installed by network operators. Their deployment is challenging because they have specific chaining requirements. As opposed to traditional physical appliances, virtual network functions (VNFs) can be dynamically deployed and reconfigured on demand, posing strict management challenges to networked systems. The selection of the most appropriate VNFs to achieve a particular objective, the decision on where to deploy these VNFs and through which paths they will communicate are the responsibilities of an NFV orchestrator. In this dissertation, we propose to orchestrate VNFs using interacting cognitive components structured with the belief-desire-intention (BDI) architecture, leading to emergent solutions to address network challenges. The BDI architecture includes a reasoning cycle, which provides agents with rational behaviour, allowing agents to deal with different scenarios in which flexible and intelligent behaviour is needed. We extend the NFV architecture, replacing its centralised orchestrator with BDI agents. Our proposal includes a reverse auction protocol and a novel bidding heuristic that allow agents to make decisions regarding the orchestration tasks. Finally, we provide a testbed that integrates a platform for developing BDI agents with a network emulator, allowing agents to control VNFs and perceive the network. This testbed is used to implement VNFs and empirically evaluate our theoretical model in a distributed denial-of-service (DDoS) attack. The evaluation results show that a solution to the DDoS attack emerges through the negotiation of agents, successfully mitigating the attack.

Multi-agent systems are a new computational approach for solving real world, dynamic and open system problems. Problems are conceptualized as a collection of decentralised autonomous agents that collaborate to reach the overall solution. Because of the agents autonomy, their limited rationality, and the distributed nature of most real world problems, the key issue in multi-agent system research is how to model interactions between agents. Negotiation models have emerged as suitable candidates to solve this interaction problem due to their decentralised nature, emphasis on mutual selection of an action, and the prevalence of negotiation in real social systems. The central problem addressed in this thesis is the design and engineering of a negotiation model for autonomous agents for sharing tasks and/or resources. To solve this problem a negotiation protocol and a set of deliberation mechanisms are presented which together coordinate the actions of a multiple agent system. In more detail, the negotiation protocol constrains the action selection problem solving of the agents through the use of normative rules of interaction. These rules temporally order, according to the agents' roles, communication utterances by specifying both who can say what, as well as when. Specifically, the presented protocol is a repeated, sequential model where offers are iteratively exchanged. Under this protocol, agents are assumed to be fully committed to their utterances and utterances are private between the two agents. The protocol is distributed, symmetric, supports bi and/or multi-agent negotiation as well as distributive and integrative negotiation. In addition to coordinating the agent interactions through normative rules, a set of mechanisms are presented that coordinate the deliberation process of the agents during the ongoing negotiation. Whereas the protocol normatively describes the orderings of actions, the mechanisms describe the possible set of agent strategies in using the protocol. These strategies are captured by a negotiation architecture that is composed of responsive and deliberative decision mechanisms. Decision making with the former mechanism is based on a linear combination of simple functions called tactics, which manipulate the utility of deals. The latter mechanisms are subdivided into trade-off and issue manipulation mechanisms. The trade-off mechanism generates offers that manipulate the value, rather than the overall utility, of the offer. The issue manipulation mechanism aims to increase the likelihood of an agreement by adding and removing issues into the negotiation set. When taken together, these mechanisms represent a continuum of possible decision making capabilities: ranging from behaviours that exhibit greater awareness of environmental resources and less to solution quality, to behaviours that attempt to acquire a given solution quality independently of the resource consumption. The protocol and mechanisms are empirically evaluated and have been applied to real world task distribution problems in the domains of business process management and telecommunication management. The main contribution and novelty of this research are: i) a domain independent computational model of negotiation that agents can use to support a wide variety of decision making strategies, ii) an empirical evaluation of the negotiation model for a given agent architecture in a number of different negotiation environments, and iii) the application of the developed model to a number of target domains. An increased strategy set is needed because the developed protocol is less restrictive and less constrained than the traditional ones, thus supporting development of strategic interaction models that belong more to open systems. Furthermore, because of the combination of the large number of environmental possibilities and the size of the set of possible strategies, the model has been empirically investigated to evaluate the success of strategies in different environments. These experiments have facilitated the development of general guidelines that can be used by designers interested in developing strategic negotiating agents. The developed model is grounded from the requirement considerations from both the business process management and telecommunication application domains. It has also been successfully applied to five other real world scenarios.

Lakatos outlined a theory of mathematical discovery and justification, which suggests ways in which concepts, conjectures and proofs gradually evolve via interaction between mathematicians. Different mathematicians may have different interpretations of a conjecture, examples or counterexamples of it, and beliefs regarding its value or theoremhood. Through discussion, concepts are refined and conjectures and proofs modified. We hypothesise that: (i) it is possible to computationally represent Lakatos's theory, and (ii) it is useful to do so. In order to test our hypotheses we have developed a computational model of his theory. Our model is a multiagent dialogue system. Each agent has a copy of a pre-existing theory formation system, which can form concepts and make conjectures which empirically hold for the objects of interest supplied. Distributing the objects of interest between agents means that they form different theories, which they communicate to each other. Agents then find counterexamples and use methods identified by Lakatos to suggest modifications to conjectures, concept definitions and proofs. Our main aim is to provide a computational reading of Lakatos's theory, by interpreting it as a series of algorithms and implementing these algorithms as a computer program. This is the first systematic automated realisation of Lakatos's theory. We contribute to the computational philosophy of science by interpreting, clarifying and extending his theory. We also contribute by evaluating his theory, using our model to test hypotheses about it, and evaluating our extended computational theory on the basis of criteria proposed by several theorists. A further contribution is to automated theory formation and automated theorem proving. The process of refining conjectures, proofs and concept definitions requires a flexibility which is inherently useful in fields which handle ill-specified problems, such as theory formation. Similarly, the ability to automatically modify an open conjecture into one which can be proved, is a valuable contribution to automated theorem proving.

Social intelligence has a huge impact on the determination of human behaviour in the society. The use of norms can contribute to advances in this social intelligence by the provision of appropriate behaviour based upon the understanding of social situations. Hence, the domain of virtual characters research has given much attention to take advantage of these characteristics of norms particularly in engineering human-like behaviour. However, a lack of capability in reasoning about norms as well as a lack of norm autonomy in virtual characters have significantly diminished the naturalism in virtual characters behaviour. Within this context, a hybrid approach incorporating social and individual reasoning inspired by socio-cognitive theory is taken into account in this thesis. To this end, we propose DNA 3 , Distributed Norm Aware Agent Architecture, established through the integration of (i) the institution, a normative framework performing the social reasoning, (ii) N-Jason, a (BDI-type) cognitive agent carrying out run-time norm-aware deliberation and (iii) a virtual character in charge of perception and realisation of actions. The institution takes responsibility of (i) analysis of state of external worlds by recording a sequence of event occurrences observed by multiple virtual agents, (ii) reasoning about situationally appropriate behaviour with an assistance from Answer Set Programming (ASP) solver depending upon the social context virtual characters encounter and (iii) in turn detachment of a new set of norms, more precisely normative consequences of specific actions, to virtual characters. This contributes to the enhancement in the flexibility in specifying and reasoning about social norms subject to changes of social situations. Those detached norms are involved in the reasoning process of in- dividual virtual characters. In here, a norm-aware BDI-type agent, N-Jason, performs a practical reasoning to select a plan to execute between norms and goals. Basically, N-Jason offers a generic norm execution mechanism on top of norm aware deliberation to contribute to the exploitation of run-time norm compliance. The selection of agent behaviour is achieved in the norm-aware deliberation process by intention scheduling with deadlines and priorities. This improves the rationality in the choice of behaviour with taking into account the preference on norms and goals in agent mind by evaluation of the importance and imminence between feasible plans triggered by both norms and goals. The design and simulation of politeness is presented as an evaluation of DNA 3 with respect to the effectiveness and adequacy in modelling virtual characters behaviour. The emphasis in here lies on the capability that is able to exhibit different types of appropriate polite behaviour in response to frequent changes in social situations. This is mainly driven by two main activities: prediction of other participants’ intention is carried out by norm-aware virtual characters whilst the understanding of context and reasoning about relevant social behaviour is performed in normative frameworks. For this purpose, three case studies are provided in this thesis: (i) politeness in navigation of individuals, (ii) politeness in the formation and navigation of groups during a guided tour, and(iii) evacuation model as a politeness in the emergency situation. The evaluation is conducted by measuring: (i) the appropriateness of in response to scenarios (e.g.a number of avoiding collisions) and (ii) the reliability of agent decision making (e.g. a response time in relation to norms with the highest priority and the most urgent).