Mehmet Sagbas
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Mehmet Sagbas has a total of 49 co-authors. Here are the top co-authors by number of shared publications:
- Umut Engin Ayten
Yildiz Technical University
20 shared publications - Shahram Minaei
Doğus University
17 shared publications - Norbert Herencsar
Department of Telecommunications, Faculty of Electrical Engineering and Communication, Brno University of Technology, Technicka 3082/12 Brno, Czech Republic
6 shared publications Umut Engin Ayten
6 shared publicationsHerman Sedef
5 shared publications- Jaroslav Koton
Brno University of Technology
4 shared publications
Latest Publications
- Open Access
A Quantum Memristor-Modulated Hopfield Chaotic System for Secure and Real-Time Image Encryption in IoT Edge Devices
This paper presents a hybrid image encryption framework based on a differential photonic quantum memristor-modulated Hopfield neural network (D-PQM–HNN). The four-dimensional D-PQM–HNN exhibits rich and complex chaotic dynamics with a borderline hyperchaotic tendency. The network is used as the entropy source of the framework, and the bitstream generated from the network passes the National Institute of Standards and Technology (NIST) SP 800-22 statistical test suite at the adopted significance level. The bitstream drives a multilayer encryption pipeline comprising bidirectional cumulative XOR diffusion, permutation-based confusion, an extended 16-symbol DNA-domain layer with rule-selected mixing, multi-point crossover and mutation, and a terminal Advanced Encryption Standard-256-in-counter-mode (AES-256-CTR) stage. The complete algorithm is implemented on an NVIDIA Jetson AGX Orin using channel-wise Compute Unified Device Architecture (CUDA) parallelization. For 512 × 512 color images, the implementation reduces the total encryption time from approximately 1.5 s on the reference desktop CPU to 0.22–0.24 s on the embedded platform. The experiments on four benchmark images include histogram and chi-square analyses, adjacent-pixel correlation, number of pixels change rate/unified average changing intensity (NPCR/UACI), information entropy, key sensitivity, and robustness against noise and data loss. The results show near-uniform cipher image statistics, high plaintext and key sensitivity, and recoverable image information under the tested corruption conditions, while demonstrating the feasibility of the proposed framework for edge-oriented image encryption.
Open AccessA Quantum Memristor-Modulated Hopfield Chaotic System for Secure and Real-Time Image Encryption in IoT Edge Devices
This paper presents a hybrid image encryption framework based on a differential photonic quantum memristor-modulated Hopfield neural network (D-PQM–HNN). The four-dimensional D-PQM–HNN exhibits rich and complex chaotic dynamics with a borderline hyperchaotic tendency. The network is used as the entropy source of the framework, and the bitstream generated from the network passes the National Institute of Standards and Technology (NIST) SP 800-22 statistical test suite at the adopted significance level. The bitstream drives a multilayer encryption pipeline comprising bidirectional cumulative XOR diffusion, permutation-based confusion, an extended 16-symbol DNA-domain layer with rule-selected mixing, multi-point crossover and mutation, and a terminal Advanced Encryption Standard-256-in-counter-mode (AES-256-CTR) stage. The complete algorithm is implemented on an NVIDIA Jetson AGX Orin using channel-wise Compute Unified Device Architecture (CUDA) parallelization. For 512 × 512 color images, the implementation reduces the total encryption time from approximately 1.5 s on the reference desktop CPU to 0.22–0.24 s on the embedded platform. The experiments on four benchmark images include histogram and chi-square analyses, adjacent-pixel correlation, number of pixels change rate/unified average changing intensity (NPCR/UACI), information entropy, key sensitivity, and robustness against noise and data loss. The results show near-uniform cipher image statistics, high plaintext and key sensitivity, and recoverable image information under the tested corruption conditions, while demonstrating the feasibility of the proposed framework for edge-oriented image encryption.
- 2 November 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (44) , 445201
Order-tunable chaotic dynamics, memory-state-dependent coexistence, and reservoir computing in a fractional memfractor-based oscillator
In this paper, an autonomous oscillator utilizing a memfractor with fractional-order elements is presented and analyzed with regard to its order-dependent chaotic behavior, memory state-dependent coexistence, and reservoir computing properties. The design includes a Lorenz-like nonlinear element and a memfractor branch where the effective input is generated using the memory state along with a fractional-order accumulation state. The memfractor is first analyzed on an elemental basis through frequency-dependent pinched v–i hysteresis loops, which confirm the expected fading-memory signature. Thereafter, the generated oscillator is analyzed using equilibria, stability metrics, phase planes, bifurcations, finite-time Lyapunov analysis, and initial-condition scans. For the default parameter set, nine equilibrium points are detected, including two locally stable equilibria and seven unstable equilibria. Under the same parameters, the system exhibits coexistence between stable equilibrium convergence and bounded chaotic oscillations, mainly selected by the initial memfractor memory state. The memfractor coupling strength controls the transition toward chaos, while the fractional orders tune the intensity of memory-induced divergence. The numerical credibility of the reported regimes is supported by Grünwald–Letnikov and predict–evaluate–correct–evaluate / Adams–Bashforth–Moulton solver cross-validation, together with time-step, simulation-length, and memory-length sensitivity tests. Finally, the oscillator is evaluated as a memory-tunable reservoir on NARMA10 and NARMA20 tasks. In the reservoir-computing tests, the full fractional memfractor configuration gives the lowest representative NARMA10 and NARMA20 prediction errors. The ablation results further indicate that fractional dynamics and memfractor feedback contribute differently to temporal retention and nonlinear prediction. These results show that the fractional memfractor provides a compact mechanism for memory-controlled chaos, coexistence, and reservoir-performance tuning.
- 2 November 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (44) , 445201
Order-tunable chaotic dynamics, memory-state-dependent coexistence, and reservoir computing in a fractional memfractor-based oscillator
In this paper, an autonomous oscillator utilizing a memfractor with fractional-order elements is presented and analyzed with regard to its order-dependent chaotic behavior, memory state-dependent coexistence, and reservoir computing properties. The design includes a Lorenz-like nonlinear element and a memfractor branch where the effective input is generated using the memory state along with a fractional-order accumulation state. The memfractor is first analyzed on an elemental basis through frequency-dependent pinched v–i hysteresis loops, which confirm the expected fading-memory signature. Thereafter, the generated oscillator is analyzed using equilibria, stability metrics, phase planes, bifurcations, finite-time Lyapunov analysis, and initial-condition scans. For the default parameter set, nine equilibrium points are detected, including two locally stable equilibria and seven unstable equilibria. Under the same parameters, the system exhibits coexistence between stable equilibrium convergence and bounded chaotic oscillations, mainly selected by the initial memfractor memory state. The memfractor coupling strength controls the transition toward chaos, while the fractional orders tune the intensity of memory-induced divergence. The numerical credibility of the reported regimes is supported by Grünwald–Letnikov and predict–evaluate–correct–evaluate / Adams–Bashforth–Moulton solver cross-validation, together with time-step, simulation-length, and memory-length sensitivity tests. Finally, the oscillator is evaluated as a memory-tunable reservoir on NARMA10 and NARMA20 tasks. In the reservoir-computing tests, the full fractional memfractor configuration gives the lowest representative NARMA10 and NARMA20 prediction errors. The ablation results further indicate that fractional dynamics and memfractor feedback contribute differently to temporal retention and nonlinear prediction. These results show that the fractional memfractor provides a compact mechanism for memory-controlled chaos, coexistence, and reservoir-performance tuning.
- Open Access
- 20 July 2026
- Published by Omer Halisdemir Universitesi in Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi
- Vol. 18
Techno-economic and environmental assessment of a 300-kWe rooftop solar power plant under industrial tariff structures and dynamic market conditions in Türkiye
This study presents a probabilistic techno-economic and environmental assessment of a 300-kWe grid-connected rooftop PV system in Manisa, Türkiye, utilizing an optimized east–west orientation. Validated through a three-tier framework that includes field data, the system generates 543.3 MWh annually, with a performance ratio of 78.99%. Crucially, it provides 18% higher energy during evening peaks compared to south-facing alternatives. Monte Carlo simulations (10.000 iterations) confirm financial robustness, yielding a mean probabilistic payback of 2.266 years, an internal rate of return of 51.04%, and a 100% probability of positive NPV. Additionally, the system avoids 255 tons of annual CO₂ emissions. These results position east–west–oriented PV systems as cost-effective, grid-friendly solutions for emerging economies, supporting Sustainable Development Goals 7, 9, and 13.
- 20 July 2026
- Published by Omer Halisdemir Universitesi in Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi
- Vol. 18
Open AccessTechno-economic and environmental assessment of a 300-kWe rooftop solar power plant under industrial tariff structures and dynamic market conditions in Türkiye
This study presents a probabilistic techno-economic and environmental assessment of a 300-kWe grid-connected rooftop PV system in Manisa, Türkiye, utilizing an optimized east–west orientation. Validated through a three-tier framework that includes field data, the system generates 543.3 MWh annually, with a performance ratio of 78.99%. Crucially, it provides 18% higher energy during evening peaks compared to south-facing alternatives. Monte Carlo simulations (10.000 iterations) confirm financial robustness, yielding a mean probabilistic payback of 2.266 years, an internal rate of return of 51.04%, and a 100% probability of positive NPV. Additionally, the system avoids 255 tons of annual CO₂ emissions. These results position east–west–oriented PV systems as cost-effective, grid-friendly solutions for emerging economies, supporting Sustainable Development Goals 7, 9, and 13.
- 24 June 2026
- Published by Elsevier in Integration
- Vol. 110
Embedding memory into a fundamental analog building block: The memOTA concept
- 24 June 2026
- Published by Elsevier in Integration
- Vol. 110
Embedding memory into a fundamental analog building block: The memOTA concept
- 20 June 2026
- Published by Elsevier in Chaos, Solitons & Fractals
- Vol. 210
Differential photonic quantum memristor–modulated hopfield neural network with secure IoMT federated learning
- 20 June 2026
- Published by Elsevier in Chaos, Solitons & Fractals
- Vol. 210
Differential photonic quantum memristor–modulated hopfield neural network with secure IoMT federated learning
- Open Access
- 4 June 2026
- Published by Springer Nature in Journal of Computational Electronics
- Vol. 25 (3) , 121
A compact memristive quantum FET (memQFET) model with Landauer transport and WKB tunneling for circuit-level simulation
This paper introduces a physics-based compact model of a three-terminal memristive quantum field-effect transistor (memQFET), which can be used for circuit-level simulation. A memory state is incorporated into the model within the Landauer transport formulation. The current is calculated through the Landauer energy integral, while the transmission probability is obtained using the Wentzel–Kramers–Brillouin (WKB) approximation of a localized barrier affected by the state variable. When the drain terminal is excited by a bipolar voltage, the time-varying memory state ( w ( t )) modulates the barrier height, resulting in the formation of pinched hysteresis loops (PHLs) in the current–voltage characteristics without the need for additional window functions or corrective terms. The model also allows observation of the internal state while preserving proper three-terminal device operation. The model has been implemented in the SPICE environment.
- 4 June 2026
- Published by Springer Nature in Journal of Computational Electronics
- Vol. 25 (3) , 121
Open AccessA compact memristive quantum FET (memQFET) model with Landauer transport and WKB tunneling for circuit-level simulation
This paper introduces a physics-based compact model of a three-terminal memristive quantum field-effect transistor (memQFET), which can be used for circuit-level simulation. A memory state is incorporated into the model within the Landauer transport formulation. The current is calculated through the Landauer energy integral, while the transmission probability is obtained using the Wentzel–Kramers–Brillouin (WKB) approximation of a localized barrier affected by the state variable. When the drain terminal is excited by a bipolar voltage, the time-varying memory state ( w ( t )) modulates the barrier height, resulting in the formation of pinched hysteresis loops (PHLs) in the current–voltage characteristics without the need for additional window functions or corrective terms. The model also allows observation of the internal state while preserving proper three-terminal device operation. The model has been implemented in the SPICE environment.
- 7 May 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (21)
High-Order Memristor-Based (4+n)-Dimensional Hyperchaotic System with Microcontroller Realization and Chaos-Assisted ECG Encryption
The rapid expansion of the Internet of Medical Things (IoMT) necessitates lightweight and robust security mechanisms for protecting sensitive biomedical signals during transmission. While memristor-based chaotic systems have emerged as promising entropy sources for cryptographic applications, their reliance on conventional first-order memory often limits their dynamical complexity. To address this limitation, this paper proposes a high-order memristor-based (4ⓜ+n)-dimensional hyperchaotic system. The memductance of the proposed hyperchaotic system depends on an n-stage internal memory chain that generalizes first-order memristive dynamics and enables richer routes to instability. The proposed (n+4)-dimensional framework is formulated for a general nand explicitly demonstrated for n=1 (5D) and n=2 (6D), corresponding to fifth- and sixth-order systems, respectively. The system’s fundamental properties are established through equilibria, dissipativity, bifurcation studies, and Lyapunov spectrum computation, yielding a Kaplan–Yorke dimension of D_KY≈5.05 for the n=2 case. In addition, the effect of the memory order n on the attractor geometry is investigated. To validate practicality, the proposed model is digitally emulated on an STM32H7A3ZI microcontroller. Finally, a chaos-assisted biomedical protection pipeline in which the proposed dynamics provide signal-adaptive masking/permutation for electrocardiogram (ECG) streams prior to secure transport is presented. Experimental evaluations show a Miller–Madow entropy of 15.930 bits, adjacent-sample correlation reduced to 0.004, key-sensitivity with a BER of 0.501, and near-lossless plaintext recovery with PRD = 0.009% and SNR = 80.56 dB. In addition, the scheme achieves NPCR = 100.00% and UACI = 33.10%, supporting lightweight on-device confidentiality for IoMT applications.
- 7 May 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (21)
High-Order Memristor-Based (4+n)-Dimensional Hyperchaotic System with Microcontroller Realization and Chaos-Assisted ECG Encryption
The rapid expansion of the Internet of Medical Things (IoMT) necessitates lightweight and robust security mechanisms for protecting sensitive biomedical signals during transmission. While memristor-based chaotic systems have emerged as promising entropy sources for cryptographic applications, their reliance on conventional first-order memory often limits their dynamical complexity. To address this limitation, this paper proposes a high-order memristor-based (4ⓜ+n)-dimensional hyperchaotic system. The memductance of the proposed hyperchaotic system depends on an n-stage internal memory chain that generalizes first-order memristive dynamics and enables richer routes to instability. The proposed (n+4)-dimensional framework is formulated for a general nand explicitly demonstrated for n=1 (5D) and n=2 (6D), corresponding to fifth- and sixth-order systems, respectively. The system’s fundamental properties are established through equilibria, dissipativity, bifurcation studies, and Lyapunov spectrum computation, yielding a Kaplan–Yorke dimension of D_KY≈5.05 for the n=2 case. In addition, the effect of the memory order n on the attractor geometry is investigated. To validate practicality, the proposed model is digitally emulated on an STM32H7A3ZI microcontroller. Finally, a chaos-assisted biomedical protection pipeline in which the proposed dynamics provide signal-adaptive masking/permutation for electrocardiogram (ECG) streams prior to secure transport is presented. Experimental evaluations show a Miller–Madow entropy of 15.930 bits, adjacent-sample correlation reduced to 0.004, key-sensitivity with a BER of 0.501, and near-lossless plaintext recovery with PRD = 0.009% and SNR = 80.56 dB. In addition, the scheme achieves NPCR = 100.00% and UACI = 33.10%, supporting lightweight on-device confidentiality for IoMT applications.
- 25 March 2026
- Published by Elsevier in Chaos, Solitons & Fractals
- Vol. 208
Dynamic memdiode–driven Hopfield neural network for 3D CT encryption in IoMT
- 25 March 2026
- Published by Elsevier in Chaos, Solitons & Fractals
- Vol. 208
Dynamic memdiode–driven Hopfield neural network for 3D CT encryption in IoMT