Projects
Research Projects
Collaboratively Train Anomaly Detection Models in Edge Computing: (Research Fellow at SUTD)
Research and implement a federated learning framework to collaboratively train an anomaly detection (AD) model for edge devices (specifically, for multiple robots who work in different environments and belong to different cooperations). The AD model not only runs inference at edge devices, but we also implemented a federated learning framework to collaboratively learn a shared model without leaking private training datasets.
Contextual-Bandit Anomaly Detection for IoT Data in Hierarchical Edge Computing: (PhD at SUTD)
In this project, we designed an adaptive scheme to select one of the machine learning models based on the contextual information extracted from input data, to perform anomaly detection. The selection is formulated as a contextual bandit problem and is characterized by a single-step Markov decision process, with an objective of achieving high detection accuracy and low detection delay simultaneously. More details are described in our papers:
[ICDCS'20] Mao V. Ngo, Tie Luo, Hakima Chaouchi, and Tony Q.S. Quek, "Contextual-Bandit Anomaly Detection for IoT Data in Distributed Hierarchical Edge Computing," IEEE International Conference on Distributed Computing Systems (ICDCS), Demo Track, July 2020. Acceptance rate: 30% (13/43) [PDF] [Demo Video]
[AAAI'20] Mao V. Ngo, Hakima Chaouchi, Tie Luo, Tony Q.S. Quek, "Adaptive Anomaly Detection for IoT Data in Hierarchical Edge Computing," AAAI Workshop on Artificial Intelligence of Things (AIoT), 2020. [PDF]
Hierarchical mobile edge computing (HEC) architecture for mobile users: (PhD at SUTD)
We designed an HEC system that jointly supports container migration and base station handover for mobile users. Source code is available in Gitlab.
In order to evaluate with some real-world applications, we have developed an android app, and several offloaded services (e.g., face recognition based on OpenFace, object recognition based on Yolo-v3). All source codes are publicly made available as well. Some video demo: https://youtu.be/6FETIIdDqe8 and https://youtu.be/7AzQ88y7K1M
More details are described in our paper:
Soldier Body Area Networks: (Research Assistant at SUTD)
Develop dynamic resource management schemes for an SBAN system.
Develop interference model and interference management techniques for coexistence of single SBAN, multiple SBANs, and SBANs with other networks.
Develop an energy-efficient MAC protocol in SBAN communication platform, incorporating dynamic resource management and interference management techniques.
Propose context-aware scheduling for SBAN to improve energy efficiency and priority of emergency messages.
Investigate the network integration of heterogeneous wireless networking technologies to support seamless roaming and end-to-end QoS.
Investigate the effect of operating frequency band on the proposed techniques.
Industrial Projects
IPv6 for APZ system: (R&D Engineer at DEK Vietnam)
Description: APZ support various types of IP stack implementations for different purposes. There is an IP stack on the Central Processor for system internal communication to I/O system for communicating to the Operation Support System (OSS). This project is to implement the IPv6 feature for APZ system as one of the network layer protocol (layer 3) on AXE platform to fully comply with all features from IPv4 connectivity.
Improved skills/knowledge: TCP/IP knowledge, networking program, C/C++.
APZ’s CP Design: (R&D Engineer at DEK Vietnam)
Design Central Processor (CP) software blocks in APZ in AXE telephone exchange system. (AXE based equipment are being used as BSC/TRC, MSC, HLR, SCP, FNR, TSC, STP and wireline nodes.)
Writing in PLEX-C language and ASA-C.