MSc Dissertation

MSc. dissertation works supervised by Prof. Joberto Martins recently.

The growing number of Internet of Things enabled applications is a trend that keeps growing. Traffic management and Quality of Servic (QoS) is a subject often researched, and is growing into the IoT  with customized mechanisms and applications, tailored to fit IoT use cases and characteristics. In this work, a Publish/Subscribe QoS-aware framework (PSIoT-Orch) that orchestrates IoT traffic and mediates the allocation of network resources between IoT data aggregators and pub/sub consumers. The PSIo-Orch relies on IoT characteristics for QoS control and network integration for better QoS capabilities. The PSIoT-Orch framework is validated in an SDN-based network and integrated into network-level QoS via BAM. The results show proper bandwidth distribution and higher link utilization when running the PSIoT-Orch. 
Pedro Francesco Dias Moraes
UNIFACS - PPGCOMP -2018
The Internet of Things (IoT) is considered a major trend in computing and in specific areas such as Smart Cities, Smart Grid, Industry 4.0 and mobile applications based on 5G. Typically, this set of technologies requires orchestration of heterogeneous resources that are allocated over distinct infrastructures such as Cloud Computing, Cloud of Things, Datacenters, and network backbones. Consistent with this demand, the PSIoT-Orch framework was designed to orchestrate massive IoT traffic and to allocate network resources between Aggregators and Consumers in a Publish / Subscribe strategy. This dissertation aims to build an intelligent module for PSIoT-Orch that is capable of handling data types with different transmission requirements, aiming at an efficient use of a limited communication link. The proposed component uses Reinforcement Learning, more specifically, the SARSA algorithm to dynamically adjust the available bandwidth according to transmission priority. This solution, named PSIoT-SARSA, is validated in a simulation environment under the statistical methods of Analysis of Variance and Response Surface Analysis and, at the end of the study, it is observed that it obtained promising results. The contributions are focused on gathering an approach that allows to allocate bandwidth in an intelligent way, allowing an efficient scheduling of the IoT flow, in the scenario of the Smart Grid. 
Carlos Eduardo Arruda de Souza
UNIFACS - PPGCOMP -2019
The popularization of computer networks, especially the Internet, with the use of services that require large data flows, has generated a growing demand for computational resources, mainly bandwidth. Bandwidth Allocation Models (BAM) has proven to be a viable alternative to network management where the bandwidth resource is shared to meet high demand for network. However, managing these networks has become an increasingly complex task and solutions that allows for nearly autonomous configuration with less intervention of the network manager are highly demanded. The use of Case-Based Reasoning (CBR) techniques has proven to be satisfactory for decision making and it is used for network management. This work presents a proposal for network reconfiguration based on the CBR cycle, intelligence and cognitive module for MPLS (Multi-Protocol Label Switching) networks. The results show that the use of CBR is a feasible solution for autoconfiguration with autonomic characteristics in the MPLS using bandwidth allocation models (BAMs). The proposal improved the network general performance. 
Eliseu Morais de Oliveira
UNIFACS - PPGCOMP -2019
Smart grids (SGs) have as one of their basic proposals to incorporate intelligence into the electric grid through computing and communication technologies aiming at greater efficiency and effectiveness in their operation and control. Power loss, quality and failures are inherent in the generation process, transmission and distribution of electricity and in the context of SGs, should be minimized to ensure greater resilience and system efficiency. Dynamic and efficient reconfiguration of the distribution network is an example of an SG functionality. The reconfiguration process consists essentially of adjusting or changing the topology of the distribution network from the opening and closing of switches in order to minimize technical losses, optimize operating parameters and restore power supply in contingency situations. The nature of the network reconfiguration problem is combinatorial, complex, and non-linear. Aiming to minimize the time of convergence in the search of solution in medium and large topologies, heuristic and optimization techniques are an alternative. This dissertation proposes a new genetic algorithm GAEnhanced (Genetic Algorithm Enhanced) to solve the problem of network reconfiguration and makes a comparative study of performance aspects of this algorithm in relation to other solutions and algorithmic strategies used. The main goal is to evaluate the algorithm implementation strategies for dynamic reconfiguration and on-thefly of distribution networks from a broader perspective, in addition to proposing a new solution with the GAEnhanced algorithm. A simulator (DNRSim) with basic functionalities for implementation and tests of network reconfiguration algorithms for the Smart Grid was also developed within the scope of this dissertation. The comparative study of the performance of the GAEnhanced algorithm and other solutions with the DNRSim uses the IEEE models for system tests (14-bus, 30-bus, 57-bus, 118-bus and 330-bus). The result of the comparative study illustrates the different ways to efficiently compute network reconfiguration solutions (scalability, time and quality) and, also demonstrates the feasibility of using the GAEnhanced algorithm in the context of Smart Grids in a perspective of deploying more autonomic and intelligent solutions. 
Alysson Rômulo de Souza Pezzutti
UNIFACS - PPGCOMP -2018
The flow of data across the Internet has grown considerably in the last years and maintains this trend in volume, complexity and variety, and therefore needs to be adequately managed by ISPs. Network neutrality represents the concept that Internet traffic in general should not be discriminated by ISPs based on its source, destination and content. To achieve this goal, a number of countries adopt a hard stance and prohibit certain abusive conducts. An open issue and challenging task is to propose an operational and effective traffic management model that can implement and comply with the concept of network neutrality on ISPs networks. This dissertation presents evidence on the use of Bandwidth Allocation Models (BAMs) as a mechanism to implement neutrality in IP/MPLS networks. BAMs, in brief, can allocate resources on demand and have distinct “behaviors” that can be exploited to achieve network neutrality. This work introduces the AllocTC-Sharing behavior reproduced by the Generalized Bandwidth Allocation Model (G-BAM), discusses how this behavior can achieve a network operation in accordance with network neutrality even using service differentiation and, finally, simulates and evaluates the compliance with the established rules, considering a non-discriminatory mapping of applications in different Traffic Classes. The results indicate that the AllocTC-Sharing behavior is appropriate and compatible with the imposed net neutrality rules.
David Santana e Silva Barreto
UNIFACS - PPGCOMP -2017
Electrical networks are composed of stages of generation, transmission and distribution of energy. Distribution networks (RD) are an important element of the electricity grid because it provides the effective delivery of energy to end users. The RD’s are subject to failure and their optimization is of fundamental importance in the context of Smart Grids, where it is sought a greater efficiency of the processes involved between the production and distribution of energy. The distribution networks (RD) have varied topologies and loads. This dissertation proposes a method and algorithm for the reconfiguration of electrical network using machine learning with linear regression and branch exchange algorithm aiming the optimization of RD operation. The method and algorithm proposed do maneuvers in the RD as transfer and load balancing, aiming at to increase its level of reliability. The proposal is validated in a test network of IEEE (IEEEbus14) using a simulation and testings environment implemented in “R” language and using the Newton Raphson method to calculate the power flow. The solution developed was satisfactory in supporting the decision-making process for there reconfiguration of distribution networks in the Smart Grid context.
Eonassis Oliveira Santos
UNIFACS - PPGCOMP -2017
Software Defined Networking (SDN) paradimg decouples control plane and of data plane, providing high programmability and a global view of the network. The adoption of this approach is growing in business networks, data centers and critical infrastructures such as smart grids. However, it is a challenge not only to provide security in these new generation networks but also to allow a network attack to be susceptible to an incident handling and forensic expertise procedure. In this way, this master’s degree dissertation proposes the implementation of a Environment for Flexible Attacks Detection and Prevention in OpenFlow/SDN Networks, which includes the setup of a real OpenFlow/SDN Testbed environment and implemented mechanisms for detection and response to threats capable of providing resources for intrusion and attack analysis. These mechanisms were implemented with the goal of providing security event monitoring and treatment in a flexible way, by categorizing the attack types and associated with whitelist and blacklist resources, exploiting one of the main characteristics of the OpenFlow network controller to be Extensible and programmable. The validation of the environment was done through simulation based on 5 (five) different scenarios and the obtained results demonstrate that the different classes of network attacks were identified and treated according to the defined strategy. Also, the mechanisms of protection and analysis of the intrusion into the OpenFlow/SDN network Testbed environment were effective and had the expected behavior according to each class of traffic defined in the proposal.
Maxli Barroso Campos
UNIFACS - PPGCOMP -2017