Horário: 14h
Palestrante:Aurélio Ribeiro Costa (doutorando)
Orientadora:Profa Célia Ghedini Ralha
Título:Towards Modularity Optimization Using Reinforcement Learning for Community Detection in Dynamic Social Networks
Resumo:The identification of community structure in social network is an important problem tackled in literature of network analysis. Many solutions to the community detection problem are constrained to static scenarios and some of them can be adapted to dynamic scenarios with limitations on the resulting performance, on the other hand, some solutions employed to static scenarios simply do not fit to dynamic scenarios, moreover when considering the demand to analyze constantly growing networks. In this context, we propose an approach to the problem of community detection in dynamic networks based on reinforcement learning strategy to deal with temporal aspects on big networks using a local optimization on modularity function of the changed entities. An experiment using synthetic and real-world dynamic network data shows results comparable to static scenarios.
Horário: 14h20
Palestrante:Aldo Henrique Dias Mendees (doutorando)
Orientadora: Profa Célia Ghedini Ralha
Título:An Autonomous Multiagent Architecture for Dynamic and Optimazed Cloud Resource Management
Resumo: Dynamic cloud resource management is a challenging issue. Thus, multi-agent system technology can offer noticeable improvements, with intelligent agents to make decisions about how much resources are needed for a virtual machine to perform a given task in the shortest time, with the lowest cost and with the least waste. Thus, this article introduces MAS-Cloud+, a multi-agent system for monitoring, predicting, and provisioning cloud computing resources. The agents use a hybrid model of rationality to make decisions, having with rules of inference, a metaheuristic model, and a model with combinatorial optimization. It instantiates virtual machines considering SLA and QoS on computing cloud platforms (i.e., Amazon EC2), prioritizing user needs in terms of time and cost to perform a task. MAS-Cloud+ was evaluated with a DNA sequencing application, where it was subjected to different sizes of workloads, which became a challenging evaluation. The choice and provisioning of virtual machines were done using the three rationality models, that is, it was possible to compare and validate the choices. And the optimized model costs 17% less when compared to the non-optimized model (inference rules).
Horário: 14h40
Palestrante:Charles Antonio Nascimento Costa (mestrando)
Orientadora:Profa Célia Ghedini Ralha
Título:An Agent-based Model Applying Trust and Reputation Over Unknown Partners for Live Video Distributed Transcoding in Open Environments
Resumo: Adaptive Bitrate Streaming (ABR) is a popular technique for providing video media over the Internet. In ABR, the streaming provider splits the video stream into small segments then transcodes them in many different bitrates. So, players can adapt to unstable network parameters minimizing interruptions on playback. However, the computational cost of transcoding a video in many formats can limit its application on live video streaming. Besides, the network overhead of transmitting simultaneously many versions of the same content is a problem. Offloading the transcoding burden to the edge of the network can deal with the problem. Users and providers of live video could benefit from a joint scheme that allowed edge devices to do the transcoding with tolerable latency and delay. This work presents an agent-based model to deal with the problem of distributed transcoding on fog-edge computing. We present an agent-based model which consists of three well-defined roles. The transcoder role is for the agents in the edge interested in receiving transcoding tasks. The viewer proxy role is for those software agents who will act for the sake of the viewers. The broker role is for the agents who will coordinate the tasks for the benefit of the other two. This work proposes combing utility functions that account for Quality of Service (QoS) with Trust and Reputation Models and Multi-armed Bandits algorithms. For doing so, we present two algorithms designed to online select the best edge nodes to perform the transcoding tasks. The algorithms are ReNoS and ReNoS-II, named from Reputation-based Node Selection. The conducted experiments indicate that the proposed approach can afford utility gain keeping good QoS as perceived by viewers, besides offering protection against fake feedback attacks delivered by malicious transcoders and viewers. The outcomes indicate that our approach can potentially be applied in real-world edge computing environments.
Horário: 15h
Palestrante:Gustavo Caltabiano Eichler (mestrando)
Orientadores: Prof. Marcelo Marotta e Profa Célia Ghedini Ralha
Título:An Agent-Based Communication Architecture to Hybrid NOMA-OMA Decisions
Resumo: The emerging of Internet of Things (IoT), massive Machine Type Communication (mMTC) and Industry 4.0, will massively increase the demand for connectivity and raise Quality of Service (QoS) requirements, regarding data rates, latency, reliability, among others that exceed the limits of current mobile communication systems. To excel these limits, both Non-Orthogonal Multiple Access (NOMA) and Orthogonal Multiple Access (OMA) technologies present individual benefits and drawbacks that can be compensated by interchanging them creating communication opportunities with enhanced Spectral Efficiency (SE). To turn feasible such an interchanging, we modeled a spectrum sharing problem and propose an Agent-Based Communication Architecture to make decisions to study the detected trade-off. Through our simulations, the results highlight opportunities to interchange NOMA and OMA meeting QoS requirements with enhanced SE.
Horário: 15h20
Palestrante:Paulo Victor Gonçalves Farias (mestrando)
Orientadores: Prof. Jacir Luiz Bordim
Título: Controle de Congestão em Redes Veiculares Através de Variação de Potência de Transmissão de Mensagens
Resumo: Em redes veiculares, os nós devem enviar mensagens periódicas aos seus vizinhos com o objetivo principal de alertá-los sobre sua posição, velocidade e direção com intuito de evitar potenciais acidentes. Porém, o envio massivo dessas mensagens pode causar congestões na rede, ocasionando na falha das aplicações de segurança. Para resolver esse problema, propõe-se uma estratégia de variação de taxa de envio de mensagens e de sua potência de transmissão associada.
Local: Teams MS - Equipe PPGI-316415 Seminário, Canal Seminários 2-2021
https://teams.microsoft.com/l/channel/19%3a24b3ddb0b17b46c6be17254988b22cc1%40thread.tacv2/Semin%25C3%25A1rios%25202-2021?groupId=93b66213-b249-467a-bcbe-dcd4255edf95&tenantId=ec359ba1-630b-4d2b-b833-c8e6d48f8059
Profa Célia Ghedini Ralha (Este endereço de email está sendo protegido de spambots. Você precisa do JavaScript ativado para vê-lo.)
Coordenadora Seminários de Pós-Graduação em Informática 2-2021