PhD Thesis

PhDs thesis works supervised by Prof. Joberto Martins recently.

Network Slicing is an essential enabler technique for 5G realization. However, its usage is not restricted to this scenario, allowing better use of available resources through virtualization. In this sense, provisioning network slices as a service becomes a business opportunity and a trending research topic. The academic community brings several contributions to architectures and strategies for slice provisioning. Nevertheless, energy awareness inside such infrastructures is still an open issue. Therefore, this work focuses on filling a research gap of energy awareness in slice instance orchestration by creating a method to increase energy awareness in multi-domain network slice provisioning. Up to now, this work counts on a proposal based on the study of the state-of-the-art and Kubernetes usage for Network Slicing emulation.
Adnei Willian Donnatti
Thesis Supervisor: Prof. Dr.  Tereza Cristina de Brito Carvalho (USP)Co-supervision: Prof. Dr. Joberto Sérgio Barbosa Martins (UNIFACS)
UNIVERSIDADE DE SÃO PAULO (USP) - 2023 - ... (Work In Progress)
Competition for network resources needs to be arbitrated according to the characteristics of each application or service to provide a satisfactory, transparent and cost-effective way of use for users. Meeting resource requests dynamically, that is, without prior allocation (scheduling) of resources to applications and services, can allow sharing of resources for a better use of the network. Due to the heterogeneity of user profiles, applications and services, the dynamic attendance to resource requests tends to generate dynamic competition for resources in the networks as a whole. Bandwidth allocation models have the attributes that allow you to define application classes and control the distribution of resources between classes intuitively. These models can be altered and / or reconfigured aiming the optimization in the use of the resource in order to evolve their behaviors in line with the profile of traffic and communication and quality requirements defined for network. The definition of behavior in resource arbitrage that reflects a better efficiency of communication and quality requirements defined for network in a given traffic profile is a complex task. This complexity can make human intervention a point of failure. In general, cognitive management systems are meant to deal with complex tasks where human intervention can be a point of failure. These systems are capable of self-configuring autonomously, in response to changes or failures, according to business policies specified by the administrators. This thesis create the cognitive and dynamic orchestration of resource allocation (bandwidth), based on BAM, for MPLS / DSTE networks. To do this, we create a new bandwidth allocation model (GBAM - Generalized Bandwidth Allocation Model) that generalizes the behaviors of the classic model and allows new combinations of strategies in a configurable way at run time; and a cognitive framework (Cognitive BAM) that dynamically learns and orchestrates the behavior of GBAM in tune with the traffic profile and communication and quality requirements set for the network.
Rafael Freitas Reale
UNIFACS/ UFBA/ UEFS -2019
The reconfiguration process of the electrical grid distribution network consists of changing its topology by closing or opening interconnection keys. Such process also focuses on supporting decision making, planning and/or real-time control of the electric network operation aiming to minimizing active power losses, load balancing, fault isolation and voltage level improvements. The task of managing and making decisions to change the electrical network topology is a complex task due to the diversity of configuration possibilities. In this context, autonomic management systems are being investigated as a feasible solution to the grid reconfiguration problem. It is expcet that management human intervention can be replaced by autonomic solutions, preferably, dynamically generated. This thesis proposes the use of Case-based Reasoning (CBR) coupled with the HATSGA algorithm for the fast reconfiguration of large distribution power networks. The suitability and the scalability of the CBR-based reconfiguration strategy using HATSGA algorithm are evaluated. The evaluation indicates that the adopted HATSGA algorithm computes new reconfiguration topologies with a feasible computational time for large networks. The CBR strategy looks for managerial acceptable reconfiguration solutions at the CBR database and, as such, contributes to reduce the required number of reconfiguration computation using HATSGA. This suggests CBR can be applied with a fast reconfiguration algorithm resulting in more efficient, dynamic and cognitive grid recovery strategy. 
Flávio Galvão Calhau
UNIFACS/ UFBA/ UEFS -2019
Network services diversity, huge network size and technology heterogeneity are currently computer network characteristics that gradually induce network management to become a highly complex task for administrators. In this context, the Autonomic Management Systems (AMS) are being investigated as possible approaches capable to deal with this inherent complexity. In effect, it is expected that new autonomic systems and solutions could, among other functionalities, both configure and optimize network resources keeping network performance characteristics on a scalable way. This paper addresses, fundamentally, the scalability problem faced by current AMS implementations. In effect, it is expected that autonomic solutions could be computed, preferably, on-the-fly and, as such, can effectively substitute human intervention in network management. We argue that a network partitioning strategy can be applied to manage the scalability problem while considering a set of requirements specified by the network administrator (QoS parameters and execution time). This thesis addresses the set of basic network management requirements considered by the proposal, details the partitioning method and, finally, evaluates the related scalability issues considering a self-managed framework (AMS) in diverse scalable scenarios. Results points to the feasibility of the partitioning method in scenarios with different traffic matrix allocation and, beyond that, under new topologies resulting from failures. 
Romildo Martins da Silva Bezerra (in memoria)
UNIFACS/ UFBA/ UEFS -2012