3. Research Interests

The active research areas at TAN's Laboratory:

Time-sensitive networking (T),

AI-driven cybersecurity (A),

NextG communication networking (N),

Time-series Analysis via Network science (TAN)

Any possible collaboration is more than welcome!!!

(See Our Capabilities at this link)

Current Interests

• Artificial intelligence

• Machine learning

• Quantum-assisted communication

• Internet of Things

• Vehicular networks

• 5G wireless communications

• Edge computing, fog computing and cloud computing

• Information centric networking, software defined networking and network function virtualization

Past Interests

• Scheduling and resource allocation in stochastic networks

• Internet of things (IOT over LTE/LTE-A network, cyber-physical systems, big data, distributed

sensing and control)

• Time series analysis and dynamic factor models (stationary and non-stationary)

• Wireless communications and networking

• Cognitive radio (full-duplex communications, software defined radio architectures, protocol design,

spectrum sensing, detection and estimation)

• Statistical signal processing, compressed sensing, and compressive sampling

• Random matrix theory

AI/ML Supporting Cybersecurity

Enhancing threat visibility & monitoring => Improving threat detection
Predicting and preventing cyber attacks => 1) Real-time response to emerging threats, 2) enhancing incident response and remediation
Automated tasks =>1) Improving efficiency, 2) reducing security teams’ workload, reducing risk of human error

Possible Risk of AI/ML to Cybersecurity - AI-powered Offensive Techniques

Machine Learning Poisoning: Attackers contaminate data pool with misleading information=>incorrect learning=>failing to identify actual threats.
Evasion Techniques: Malicious actors use AI to develop malware that can change its code or behavior to evade detection by traditional antivirus software trained on static datasets.
Phishing Gets Smarter: With access to AI tools, malicious actors can mimic trusted sources=>easy to trick victims.
Deepfakes: Deepfake technology uses AI-generated images, videos and voices to mimic real people saying or doing things=> attack targeting organizations’ confidential information.
Autonomous Attack Bots: AI-enabled cyberattacks run on autopilot without human intervention => leading to fast, continuous and adaptive threats/attacks (i.e. infecting computers, carrying out cyber-attacks, stealing sensitive data, etc.).

Human action recognition for smart health monitoring

Propose joint semi-supervised classifier and deep generative model for recognizing human activities.
Develop Compressive Sensing-Variational Autoencoder (CSVAE) for video compression.
Implement distributed and centralized communication protocols for data transmissions.

EdgeAI-empowered Data Communication, Processing and Control for Military Operations

1) AI-enabled Framework for Communication and Computing.

2) AI-empowered Framework for NextG Network Slicing.

3) AI-enabled Secure, Privacy preserving and Trustworthy Computing Services.

KPIs: spectrum utilization, spectrum risks, latency, high density of connections, efficiency, agility, reliability and security.

AI aided Communication and Computing Resource Allocation to Support Blockchain-enabled Video Streaming

1. Improve operating efficiency and maximize transcoding rewards.

2. Model vehicular mobility by highly-realistic Semi-Markov renewal process.

3. Propose multi-timescale actor-critic-reinforcement learning framework to tackle challenges of dynamic variants of environment, resources, mobility & real-time video service delay.

4. Propose mobility-aware estimation for large timescale model.

5. Enhance mobility prediction performance by using analysis and classical ML.

Google Sites

Report abuse