Computer Science student (UFMG) and undergraduate researcher at the Wireless Networks Lab (WiNet/UFMG) and the CCSC Research Lab (UFMG), focused on cybersecurity, AI (reinforcement learning), and computer networks. Former Brazilian Army officer (Second Lieutenant): commanded platoon-level units in operational routines and led administrative sections, spending more time in the finance section.
Undergraduate Research (IC)
Oct 2024 — Apr 2026 (volunteer: Oct 2024 — Apr 2025 • CNPq scholar: May 2025 — Apr 2026)As unmanned aerial vehicles (UAVs) demand wireless technologies for their communication, they become prone to serious security threats that aim to compromise the integrity and confidentiality of the exchanged control data, standing out man-in-the-middle (MiM) attacks as one of the most damaging. Besides, UAV networks require essential critical services for their regular operation, like the UAV location service. Key exchange is one way to protect UAV communications against MiM attacks. This paper proposes KEYSUAV, a key exchange scheme to enable a resilient location information sharing on UAV networks against MiM attacks. KEYSUAV relies on opportunistic approaches and on the lightweight cryptographic standard ASCON, an algorithm designed for resource-constrained devices like UAVs. Simulation results show that KEYSUAV over the FlySafe system detected 100% of compromised messages sent by a MiM attacker with a slight reduction in UAVs’ spatial awareness, around 4.5%, thus fostering the resilience of the location service for UAV networks.
Publication
A. Batista, V. Trindade, and A. Santos, “An Opportunistic Key Exchange Scheme for Location Information Sharing on UAV Networks Resilient to MiM Attacks,” in Anais do XLIV Simpósio Brasileiro de Redes de Computadores e Sistemas Distribuídos, Praia do Forte/BA, 2026, pp. 141–154, doi: https://doi.org/10.5753/sbrc.2026.19890.
Supervision
Repository
Undergraduate Research (IC)
May 2026 — Apr 2027 (CNPq scholar)The safe operation of UAV networks depends on critical infrastructure systems with high levels of robustness, reliability, and security during their operation. The location sharing service is pivotal to UAVs have spatial awareness of the network, thus supporting the execution of various critical systems. However, this service is subject to man-in-the-middle (MiM) attacks that aim to disrupt its operation. This paper presents CIRRUS, a distributed mechanism in which multi-agent reinforcement learning (MARL) under centralized training and decentralized execution (CTDE) detects such attacks and identifies the attacker behind the identity of its victims, and a consensus among neighbors corroborates each identification. Over FlySafe, CIRRUS detected 99.67% of the forged location messages, identified the MiM in all simulations with 0.36% of false alarms, and held the location error at 0.50m, against 3.55m under attack.
Supervision
Course Project — Wireless Networks
Mar 2026 — Feb 2027Wireless sensor networks now carry multimedia and streaming traffic that demands high throughput. Dual-radio platforms answer this demand, and the dual-path same-parity protocol moves data over two node-disjoint paths of equal hop-count parity, so that source and destination use both radios at once, but only for a single source/destination pair. This work extends dual-path same-parity routing to multiple simultaneous source/destination pairs. We formulate the joint routing as an integer linear program with node exclusivity across pairs, including a variant that maximizes the number of admitted pairs before minimizing route cost, design greedy heuristics with different pair-ordering rules, and add a geometry-based channel assignment. The protocol is evaluated in ns-3.38 with real UDP and TCP traffic over IPv6/6LoWPAN and IEEE 802.15.4, measuring admission, goodput, delivery and delay as demand, topology and transport vary. The work started as a course project in Wireless Networks and is being prepared for publication at the professor’s invitation.
Supervision
Co-authorship
Undergraduate Final Project (TCC)
Aug 2026 — Jul 2027UAV swarms organized as FANETs rely on local consensus services, in which each UAV updates a shared state through exchanges with its neighbors; clock synchronization is the application case studied here. Because control messages carry no authentication, a man-in-the-middle (MitM) attacker that belongs to the swarm can capture a neighbor’s consensus message, alter the value it carries and reinject it under the victim’s identity, and the convergence mechanism itself then propagates the corrupted value through the whole connected component. The defense uses multi-agent reinforcement learning (MARL) under centralized training and decentralized execution (CTDE) to detect that attack and identify the UAV responsible for it: each UAV carries its own agent and decides using only the information available in its neighborhood, labeling every message so that a forged contribution is dropped before the service absorbs it, and pointing out the attacker. The service and the defense are built on ns-3.38 with ns3-ai and evaluated with 40 UAVs in a sparse, permanently partitioned topology.
Supervision
Scroll to explore →
1 / 3
Wireless Networks Lab (WiNet) & CCSC Research Lab — UFMG • Advisor: Prof. PhD Aldri Luiz dos Santos
Brazilian Army — 4th Light Mountain Signals Company
CPOR-BH
iJunior UFMG — Junior Enterprise
Federal University of Minas Gerais (UFMG)
2020 — Present
Federal Institute of Minas Gerais (IFMG)
2017 — 2019
Swipe or drag → click any card to view
Commendation from the Commander of the 4th Light Mountain Signals Company 4th Light Mountain Signals Company — 2022 
Commendation from the Commander of the 4th Light Mountain Signals Company 4th Light Mountain Signals Company — 2023 2 images 
Commendation from the Administrative Inspector of the 4th Light Mountain Signals Company 4th Light Mountain Signals Company — 2023 
Correia Lima Award Medal CPOR-BH — 2021 
1st overall in class CPOR-BH — 2021 
Best physical fitness CPOR-BH — 2021 
FIT Operation distinction (Fiber, Initiative, and Tenacity) CPOR-BH — 2021 
