Banca de DEFESA: GEORGE RAVELLY DA SILVA RIBEIRO

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : GEORGE RAVELLY DA SILVA RIBEIRO
DATE: 13/03/2026
TIME: 14:00
LOCAL: Sala Google Meet - meet.google.com/myz-yfzm-fjw
TITLE:

Integrating Edge Intelligence with Optimized AI Deployment on a Serverless Platform for Smart Cities

KEY WORDS:

Smart Cities, Edge Computing, Fog Computing, Servless Middleware.

PAGES: 87
BIG AREA: Ciências Exatas e da Terra
AREA: Ciência da Computação
SUBÁREA: Metodologia e Técnicas da Computação
SPECIALTY: Engenharia de Software
SUMMARY:

The advancement of Smart Cities depends on distributed architectures capable of realtime
data processing under dynamic workloads while operating on resource-constrained
hardware. This scenario positions Edge Intelligence (EI) as a vital strategy to decentralize
processing from the cloud, enabling Machine Learning (ML) models to run locally to
minimize latency. However, the practical application of EI remains hindered by severe
resource scarcity in IoT devices and the complexity of orchestrating model updates across
the Edge-Fog-Cloud Continuum.
This work introduces SAPPARCHI Edge AI, an evolution of a serverless middleware
platform designed to make intelligence at the edge both adaptive and resilient. The
proposed architecture addresses the AI lifecycle through three main fronts: optimized
embedded execution on low-power microcontrollers, an Over-the-Air (OTA) update system
with automatic recovery support, and a dynamic offloading mechanism powered by the
SCALE-GA genetic algorithm. This mechanism utilizes contextual metrics such as latency
and node availability to decide, in real-time, the optimal layer for task execution.
The proposal was validated using an integrated physical prototype, and expanded through
large-scale simulations in CloudSim Plus. The results confirm the feasibility of running local
inference on the ESP32 while maintaining response times suitable for critical applications.
During OTA updates, the system ensured service continuity by automatically redirecting
tasks to the Fog layer and reintegrating the edge device immediately after the process. Stress
tests revealed that the adaptive offloading naturally prioritizes the edge but proactively
adjusts to overloads. Finally, it was identified that decentralizing the control plane to
the Fog layer is a determining factor for system scalability. SAPPARCHI Edge AI thus
establishes itself as an autonomous and efficient infrastructure for distributed intelligence
in urban environments.

 

COMMITTEE MEMBERS:
Presidente - 1678918 - NELIO ALESSANDRO AZEVEDO CACHO
Interno - 3139050 - ROGER KREUTZ IMMICH
Externo ao Programa - 2669476 - ARTHUR EMANOEL CASSIO DA SILVA E SOUZA - UFRNExterna à Instituição - FLAVIA COIMBRA DELICATO - UFF