Defesa – Dissertação de Matheus Saueressig

Aluno(a): Matheus Saueressig
Orientador(a): Jéferson Campos Nobre
Coorientador(a): Thais Vasconcelos Batista

Título: Value-Based Decision and Control in Programmable Networks: Principled Prioritization of Heterogeneous Services
Linha de Pesquisa: Arquiteturas, Protocolos e Gerência de Redes e Serviço

Data: 24/03/2026
Hora: 16:00
Local: Esta banca ocorrerá de forma híbrida (virtual e presencial), na sala AUD-1 (43412) do Instituto de Informática/UFRGS e pelo link https://mconf.ufrgs.br/webconf/00094248.

Banca Examinadora:
-Alberto Egon Schaeffer Filho (UFRGS)
-Carlos Raniery Paula dos Santos (UFSM)
-Luciano Paschoal Gaspary (UFRGS)

Presidente da Banca: Jéferson Campos Nobre

Resumo: Programmable networks, such as Software-Defined Networking and programmable data planes, allow changes to forwarding, quality-of-service, and security enforcement at runtime. Nevertheless, the majority of control loops optimize narrow objectives (throughput, latency, utilization) and offer weak justification for which services to prioritize protection when constraints bind and objectives conflict. This dissertation positions network control as a value-based, multi-criteria decision problem: what to do must mirror stakeholder concerns, stick to hard safety constraints, and be explainable and reproducible. We first conduct a systematic mapping of research at the intersection of Software Engineering and programmable networks, considering the representation of objectives, artifacts (requirements, policies, telemetry, tests), and operational decisions. The mapping shows fragmentation: tradeoffs are usually implicit, constraints are inconsistently encoded and reproducibility of decision logic is infrequent – inspiring a unified decision-oriented process. We then present Dhana, a value-aware congestion-management mechanism that prioritizes high-value services under contention by adapting path selection using explicit value signals. Dhana shows how value-aware routing improves the results for critical services during overload, while exposing one central problem: shared infrastructure creates cross-service conflicts in which the benefits associated with one service may harm another. Finally, we describe Attha, a value-based decision and control system for programmable networks. For Attha, a service catalog containing service-level objectives, hard gates (non-negotiable constraints), and stakeholder weights over canonical drivers (Availability, Timeliness, Security Risk, Energy) is consumed alongside telemetry metrics. It yields reproducible per-service PASS/FAIL outcomes and a global priority ranking, and acts correctly under hard constraints and cross-service tradeoffs. In conclusion, the dissertation strengthens decision-centric ground for programmable-network operations: stakeholder-aligned control under constraints, with explainable, reproducible outputs.

Palavras-Chave: Computer Networks. Programmable Networks. Software Engineering. Value- based Software Engineering. Software Defined Network.