Banca de QUALIFICAÇÃO: DANIEL AZEVEDO DE OLIVEIRA MAIA

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DISCENTE : DANIEL AZEVEDO DE OLIVEIRA MAIA
DATA : 19/11/2026
HORA: 14:30
LOCAL: IMD 206B
TÍTULO:

CRAWO: Custom Resources for Adaptive Workload Orchestration

PALAVRAS-CHAVES:

Edge Computing; Cloud Computing; Distributed Pipelines; Resource Allocation; Self-Adaptation; MAPE-K.

PÁGINAS: 60
RESUMO:

Distributed applications in edge and cloud environments are often organized
into multiple stages with different computational and communication require-
ments. Orchestrating these applications means allocating each stage to a
processing node, at the edge or in the cloud, that is able to execute it, so
that the application achieves the required performance. This choice must
consider hardware heterogeneity, network latency, processing capacity, and
load, and must be revised during execution, since nodes may fail or become
overloaded. This work proposes CRAWO, Custom Resources for Adaptive
Workload Orchestration, an architectural framework for the adaptive orchestra-
tion of multi-stage applications over heterogeneous edge and cloud resources,
organized according to the MAPE-K feedback loop. The framework was imple-
mented and evaluated with a license plate recognition pipeline based on video,
whose detection, recognition, and notification stages have different hardware
requirements and are allocated and reallocated by CRAWO. Preliminary
experiments, in an emulated environment of heterogeneous nodes, evaluate
the time taken to select the node for each stage, the placement of stages, and
reassessment after node failure and sustained overload. The decision time
remained in the order of tenths of a millisecond per stage, growing sublinearly
from 4 to 12 candidate nodes, and the affected stages were reallocated without
moving the remaining ones.

MEMBROS DA BANCA:
Presidente - 1213777 - THAIS VASCONCELOS BATISTA
Interno - 2316877 - EVERTON RANIELLY DE SOUSA CAVALCANTE
Externo ao Programa - 2149456 - ALUÍZIO FERREIRA DA ROCHA NETO - UFRN