Martí Navarro

RT-MOVICAB-IDS

This study presents a novel Hybrid Intelligent Intrusion Detection System (IDS) known as RT-MOVICAB-IDS that incorporates temporal control. One of its main goals is to facilitate real-time Intrusion Detection, as accurate and swift responses are crucial in this field, especially if automatic abortion mechanisms are running. The formulation of this hybrid IDS combines Artificial Neural Networks (ANN) and Case-Based Reasoning (CBR) within a Multi-Agent System (MAS) to detect intrusions in dynamic computer networks. Temporal restrictions are imposed on this IDS, in order to perform real/execution time processing and assure system response predictability. Therefore, a dynamic real-time multi-agent architecture for IDS is proposed in this study, allowing the addition of predictable agents (both reactive and deliberative). In particular, two of the deliberative agents deployed in this system incorporate temporal-bounded CBR. This upgraded CBR is based on an anytime approximation, which allows the adaptation of this Artificial Intelligence paradigm to real-time requirements. Experimental results using real data sets are presented which validate the performance of this novel hybrid IDS. Research highlights? RT-MOVICAB-IDS, a novel hybrid IDS incorporating temporal control, is presented. ? Temporal restrictions allow real-time processing and system response predictability. ? Experimental results using real data sets are presented which validate the performance.

Applying dialogue games to manage recommendation in social networks

Recommendation in social networks is a new area of research that is still in its early beginnings. In this framework, every user can act as an individual recommender for its neighbours in the network. However, social networks are highly dynamic environments where the structure of the network and the information spread across it evolve quickly over time. In these settings, a suitable recommender must be able to manage continuous changes and to provide users with up-to-date and customised recommendations. With this aim, the theory of dialogue games has been applied to manage recommendation dialogues in social networks in this research. As a result, a dialogue game for controlling the interaction between an agent asking for recommendations to other personal agents that are its neighbours in the network has been designed. In addition, a complex decision-making policy based on this game has been developed and tested in a simulation scenario. The results are shown and discussed in this paper.

Incorporating temporal-bounded CBR techniques in real-time agents

Nowadays, MAS paradigm tries to move Computation to a new level of abstraction: Computation as interaction, where large complex systems are seen in terms of the services they offer, and consequently in terms of the entities or agents providing or consuming services. However, MAS technology is found to be lacking in some critical environments as real-time environments. An interaction-based vision of a real-time system involves the purchase of a responsibility by any entity or agent for the accomplishment of a required service under possibly hard or soft temporal conditions. This vision notably increases the complexity of these kinds of systems. The main problem in the architecture development of agents in real-time environments is with the deliberation process where it is difficult to integrate complex bounded deliberative processes for decision-making in a simple and efficient way. According to this, this work presents a temporal-bounded deliberative case-based behaviour as an anytime solution. More specifically, the work proposes a new temporal-bounded CBR algorithm which facilitates deliberative processes for agents in real-time environments, which need both real-time and deliberative capabilities. The paper presents too an application example for the automated management simulation of internal and external mail in a department plant. This example has allowed to evaluate the proposal investigating the performance of the system and the temporal-bounded deliberative case-based behaviour.

Real-time CBR-agent with a mixture of experts in the reuse stage to classify and detect DoS attacks

Security is a major concern when service environments are implemented. This has led to the proposal of a variety of specifications and proposals based on soft computing methods to provide the necessary security for these environments. However, most proposed approaches focus only on ensuring confidentiality and integrity, without putting forward mechanisms that ensure the availability of services and resources offered. A considerable number of attack mechanisms can lead to a web service system crash. As a result, the web service cannot allow access to authorized users. This type of attack is a so-called denial of service attack (DoS) which affects the availability of the services and recourses available. This article presents a novel soft computing-based approach to cope with DoS attacks, but unlike existing solutions, our proposal takes into account the different soft computing mechanisms that can lead to a DoS attack. Our approach is based on a real time classifier agent that incorporates a mixture of experts to choose a specific classification technique depending on the feature of the attack and the time available to solve the classification. With this scheme it is possible to divide the problem into subproblems, solving the classification of the web service requests in a more simple and effective way and always within a time bound interval. This research presents a case study to evaluate the effectiveness of the approach and also presents the preliminary results obtained with an initial prototype.

Deadline prediction scheduling based on benefits

This paper describes a scheduling algorithm that composes a scheduling plan which is able to predict the completion time of the arriving tasks. This is done by performing CPU booking. This prediction is used to establish a temporal commitment with the client that invokes the execution of the task. This kind of scheduler is very useful in scenarios where Service-Oriented Computing is deployed and the execution time is used as a parameter for QoS. This scheduler is part of an architecture that is based on the Distributed Goal-Oriented Computing paradigm, which allows agents to express their own goals and to reach them by means of service compositions. Moreover, the scheduler is also able to prioritize those tasks which provide greater benefits to the OS. In this work, the scheduler has been designed in several iterations and tested by means of a set of experiments that compare the scheduler algorithm with a representative set of scheduling algorithms. Highlights? Scheduling algorithm that composes plans which predict the completion time of tasks. ? This scheduler is part of the Distributed Goal-Oriented Computing paradigm. ? Goal-Oriented paradigm increases OS fault-tolerance and makes it adaptive. ? The scheduler prioritizes those tasks which provide greater benefits to the OS.

Towards real-time agreements

In this paper, we deal with the problem of real-time coordination with the more general approach of reaching real-time agreements in MAS. Concretely, this work proposes a real-time argumentation framework in an attempt to provide agents with the ability of engaging in argumentative dialogues and come with a solution for their underlying agreement process within a bounded period of time. The framework has been implemented and evaluated in the domain of a customer support application. Concretely, we consider a society of agents that act on behalf of a group of technicians that must solve problems in a Technology Management Centre (TMC) within a bounded time. This centre controls every process implicated in the provision of technological and customer support services to private or public organisations by means of a call centre. The contract signed between the TCM and the customer establishes penalties if the specified time is exceeded.

Distributed goal-oriented computing

For current computing frameworks, the ability to dynamically use the resources that are allocated in the network has become a key success factor. As long as the size of the network increases, it is more difficult to find how to solve the problems that the users are presenting. Users usually do know what they want to do, but they do not know how to do it. If the user knows its goals it could be easier to help him with a different approach. In this work we present a new computing paradigm based on goals. This paradigm is called Distributed goal-oriented computing paradigm. To implement this paradigm an execution framework for a goal-oriented operating system has been designed. In this paradigm users express their goals and the OS is in charge of helping the achievement of these goals by means of a service-oriented approach.

Ensuring Time in Real-Time Commitments

This paper presents a framework for the estimation of the temporal cost of agent commitments. The work here presented focuses on the development of a commitment manager as a special module in a real-time agent. This manager has been constructed for the previous analysis of a commitment request in a temporal bounded way. The proposed commitment manager incorporates as CBR module whose aim is to decide if the agent will be able to perform a requested service without exceeding a specified deadline. The proposal has been tested over a mobile robot example and this paper presents the results obtained.

Mathematical model for a temporal-bounded classifier in security environments

Security is a major concern when web applications are implemented. This has led to the proposal of a variety of specifications and approaches to provide the necessary security for these environments. SQL injection attacks on web applications have become one of the most important information security concerns over the past few years. The purpose of this article is to present an adaptive and intelligent mechanism that can handle SQL injection attacks taking into account a controlled time response. Our approach is based on a soft real-time classifier agent that incorporates a mixture of experts based on soft computing to choose a specific classification technique depending on the attack and the time available to solve the classification. A case study to evaluate the effectiveness of the approach and the preliminary results obtained with an initial prototype are also presented.

Temporal bounded reasoning in a dynamic case based planning agent for industrial environments

This paper presents a planning model integrated within a TB-CBP-BDI real-time intelligent agent that provides special abilities for planning in a predictable time, which makes its use especially appropriate in systems where certain temporal constraints must be satisfied. The proposed TB-CBP-BDI real-time agent is the core of a multi-agent system that manages security issues in industrial environments, where time constraints are a key factor. The proposed planning model facilitates the automatic temporal bounded reorganization of tasks to provide the system with adaptation abilities to the changes that occur in the environment. The planning mechanism focuses on optimizing industrial and manufacturing processes, specifically the tasks performed by the available security entities in these environments. Additionally, several Ambient Intelligence technologies such as QR-CODES, GPS, Wi-Fi and HSDPA are used to develop the intelligent environment that was tested and analyzed in this study.