Programmes & systems

Projects

Funded research and deployed engineering systems spanning networking, IoT, digital twins, smart-grid monitoring, intelligent control and wireless sensing.

8Funded programmes
6Applied systems
2008–nowSpan

Research programmes

Funded work on communication networks, from age-aware transport on terrestrial links to flow control for delay-tolerant and deep-space settings.

Fig. 02 · IAC 2026 Figure 2 from “Goal-Oriented Bundle Management and Flow Control for Deep-Space Communications via Reinforcement Learning”: Masked Lagrangian PPO architecture for event-driven relay control
Figure 2 reproduced from S. Baghaee et al., “Goal-Oriented Bundle Management and Flow Control for Deep-Space Communications via Reinforcement Learning,” 77th International Astronautical Congress (IAC 2026), 2026. View paper.
01 · European Research Council

GO-SPACE

Goal-oriented flow control for deep-space and delay-tolerant networking, where propagation delay is measured in minutes and disruption is the normal case.

  • Goal-oriented and reinforcement-learning-based flow control for disruption-prone links.
  • Analysis of timeliness–reliability trade-offs under long propagation delay.
Role
Senior Researcher
Host
METU
Period
2023–present

Latest output — Goal-Oriented Bundle Management and Flow Control for Deep-Space Communications via Reinforcement Learning, accepted for IAC 2026.

ASTRA: age-aware satellite random access Communication & networking
Fig. 03 · A³L-FEC 2024 Figure 3 from “A³L-FEC: Age-Aware Application Layer Forward Error Correction Flow Control”: Chunks in the A3L-FEC-FSFB protocol
Figure 3 reproduced from S. Baghaee and E. Uysal, “A³L-FEC: Age-Aware Application Layer Forward Error Correction Flow Control,” arXiv preprint arXiv:2410.05852 [cs.NI], 2024. View paper.
02 · TÜBİTAK

A³L-FEC — age-aware flow control

An application-layer flow controller that uses forward error correction to keep peak age below a threshold, rather than optimising throughput. The subject of a doctoral dissertation and a patent application.

  • Age-aware flow control with selective reliability over UDP.
  • Freshness–congestion trade-offs evaluated against TCP-BBR and ACP+.
Role
Ph.D. researcher
Outputs
Preprint, patent, dissertation
A³L-FEC Patent application Dissertation

Infrastructure & IoT platforms

Systems taken from hardware and embedded software through to deployment: grid monitoring in live substations, and environmental sensing feeding digital twins.

Fig. 01 · SIU 2025 Figure 1 from “AoI-Driven IoT Fault Detection for Smart Grids”: AoI-driven IoT fault detection device on a transformer
Figure 1 reproduced from S. Baghaee et al., “AoI-Driven IoT Fault Detection for Smart Grids,” 2025 33rd Signal Processing and Communications Applications Conference (SIU), 2025. View paper.
03 · Energy Market Regulatory Authority

Age-sensitive grid monitoring

AoI-aware fault indication for electricity distribution networks, taken from protocol design through hardware to pilot deployment in transformer substations, and extended with edge-AI detection in GridAI.

  • IoT fault-indicator node: hardware design, embedded software and adaptive communication.
  • Edge inference that decides locally what is worth transmitting.
Role
CTO, FreshData
Period
2023–present
AoI-driven fault detection GridAI
Fig. 02 · MipMap Digital Twin 2025 Figure 2 from “MipMap Digital Twin: A Scalable IoT-Based Framework for Smart Cities and Circular Economy”: MipMap Digital Twin Platform: Merging the physical world, real-time data, and automation for smart environments
Figure 2 reproduced from S. Baghaee and Y. Eren, “MipMap Digital Twin: A Scalable IoT-Based Framework for Smart Cities and Circular Economy,” 2025 33rd Signal Processing and Communications Applications Conference (SIU), 2025. View paper.
04 · TÜBİTAK 1004, SUIT Platform

SUIT SensorBox & digital twins

A low-power environmental sensing platform integrating heterogeneous sensors, LoRa communication, edge processing and MQTT pipelines, deployed at the AE+T Department of Delft University of Technology.

  • Occupancy-aware analytics and digital-twin-driven environmental monitoring.
  • Ontology-based smart-building semantics for portable sensor data.
Role
PI, Work Package P9
Period
2023–present
MipMap digital twin EnergyPlus–SAREF pipeline

Intelligent control & sensing

Learning applied to control problems, and sensing that reads the physical world from signals that are already in the air.

Fig. 01 · SIU 2025 Figure 1 from “Analysis of Model-Agnostic Meta-Reinforcement Learning on Automated HVAC Control”: MAML-RL framework for HVAC automation
Figure 1 reproduced from U. Filiz et al., “Analysis of Model-Agnostic Meta-Reinforcement Learning on Automated HVAC Control,” 2025 33rd Signal Processing and Communications Applications Conference (SIU), 2025. View paper.
05 · SUIT Platform & METU

Learning-based HVAC control

Reinforcement and meta-reinforcement learning for building climate control, balancing indoor air quality against energy use, with an LSTM surrogate simulator to shorten the training loop.

  • Meta-RL for transferring a policy between climates rather than retraining.
  • IoT-based multi-speed control evaluated in an instrumented room.
Role
Senior Researcher
Outputs
Five papers, 2018–2026
Meta-RL transfer learning AI, learning & control
Fig. 01 · SIU 2026 Figure 1 from “Evaluating the Impact of CSI Preprocessing on WiFi-Based Human Activity Recognition”: System design and setup for CSI data acquisition
Figure 1 reproduced from Y. C. Çelik et al., “Evaluating the Impact of CSI Preprocessing on WiFi-Based Human Activity Recognition,” 2026 34th Signal Processing and Communications Applications Conference (SIU), 2026. View paper.
06 · METU

Wi-Fi CSI sensing

Device-free human activity recognition from Wi-Fi channel state information: a microsecond-synchronised acquisition pipeline, a public dataset and a controlled deep-learning benchmark.

  • Effect of preprocessing on accuracy and latency.
  • Architecture versus model capacity across seven networks.
Host
METU
Outputs
Dataset and benchmark, 2026
REAL-CSI benchmark Wireless sensing

Industrial & applied systems

Fig. 05 · SIU 2025 Figure 5 from “Benchmarking Deep Learning Models For Automated Waste Classification”: Working Waste Segregation Device
Figure 5 reproduced from M. Dönmez et al., “Benchmarking Deep Learning Models For Automated Waste Classification,” 2025 33rd Signal Processing and Communications Applications Conference (SIU), 2025. View paper.

  • LoRa-based landfill monitoring

    LoRa-based environmental monitoring nodes deployed at a landfill site, reporting over a long-range low-power link.

    JeoIT

  • IoT-enabled smart facility management

    Real-time environmental monitoring for facility management.

    JeoIT

  • IoT-enhanced load cell system

    Recycling efficiency and unused iron plate tracking in manufacturing.

    JeoIT

  • Waste characterisation and sorting

    Capacitive and dielectric material characterisation and deep-learning waste classification for recycling.

    Material characterisation Waste classification benchmark

    METU / industry

  • Telecommunications infrastructure

    Installations of OTN centres, switch centres, rural GSM systems and power supplies at Parto Afshan Aria Co.

    2008–2010

Earlier programmes

  • AoI in future networks — optimal sampling and scheduling

    Sampling, scheduling and flow control strategies for minimising Age of Information under limited feedback.

    Age of Information publications

    Huawei

  • E-CROPS — energy-harvesting communication networks

    Communication and scheduling policies for networks that run on harvested energy, demonstrated on a vibration-powered wireless sensor testbed.

    Vibration energy-harvesting testbed

    CHIST-ERA / TÜBİTAK

  • Building Smart Energy Management System

    Q-learning-based HVAC optimisation for residential and commercial buildings, balancing indoor air quality against energy use.

    User comfort and energy efficiency by Q-learning

    TÜBİTAK 1150648

  • Magnetic wireless sensor networks

    Detection, tracking and identification of ferromagnetic targets on a magnetic sensor network testbed, from the M.Sc. thesis onward.

    Enhanced target localization

    TÜBİTAK 110E252

  • Energy-efficient design of wireless networks

    Principles and experimental implementation toward energy-efficient wireless network design.

    Wireless sensor network publications

    TÜBİTAK