A Physics-Informed Graph Neural Network Framework for Real- Time Thermal-Aware Fault Prediction and Adaptive Power Optimization in Heterogeneous System-on-Chip Architectures

Notice

This is an unedited manuscript accepted for publication and provided as an Article in Press for early access at the author’s request. The article will undergo copyediting, typesetting, and galley proof review before final publication. Please be aware that errors may be identified during production that could affect the content. All legal disclaimers of the journal apply.

Year : 2026 | Volume : 16 | 02 | Page :
By

Bibhu Prasad Ganthia,

Subash Ranjan Kabat,

  1. Assistant Professor, Department of Electrical Engineering, Indira Gandhi Institute of Technology, Sarang, Dhenkanal, Odisha, India
  2. Professor and Principal, Department of Electrical Engineering, Indira Gandhi Institute of Technology, Sarang, Dhenkanal, Odisha, India

Abstract

Heterogeneous System-on-Chip (SoC) architectures are increasingly adopted in edge computing, artificial intelligence, autonomous systems, and high-performance embedded platforms due to their superior computational efficiency and flexibility. However, increasing integration density and workload diversity introduce severe thermal hotspots, accelerated device degradation, and unexpected hardware faults that adversely affect system reliability and energy efficiency. This study proposes a Physics-Informed Graph Neural Network (PI-GNN) framework for real-time thermal- aware fault prediction and adaptive power optimization in heterogeneous SoC architectures. The proposed model integrates graph neural networks with physics-based thermal constraints to accurately capture spatial–temporal dependencies among processing elements, memory modules, and interconnects while preserving thermal consistency. Real-time sensor measurements, workload characteristics, and power profiles are fused to predict potential fault locations before failure occurrence and dynamically optimize voltage-frequency scaling, task scheduling, and power allocation. The framework minimizes thermal stress while maintaining computational performance under varying operating conditions. Extensive simulation experiments demonstrate improved fault prediction accuracy, reduced peak junction temperature, lower energy consumption, and enhanced overall system reliability compared with conventional machine learning and deep learning approaches. The proposed framework provides a scalable, intelligent, and energy-efficient solution for next-generation electronic design automation, reliable embedded computing, and advanced semiconductor system optimization across diverse heterogeneous computing environments.

Keywords: Physics-Informed Graph Neural Network (PI-GNN); Heterogeneous System-on-Chip (SoC); Thermal-Aware Fault Prediction; Adaptive Power Optimization; Electronic Design Automation (EDA); Edge Computing.

How to cite this article: Bibhu Prasad Ganthia, Subash Ranjan Kabat. A Physics-Informed Graph Neural Network Framework for Real- Time Thermal-Aware Fault Prediction and Adaptive Power Optimization in Heterogeneous System-on-Chip Architectures. Journal of VLSI Design Tools and Technology. 2026; 16(02):-.
How to cite this URL: Bibhu Prasad Ganthia, Subash Ranjan Kabat. A Physics-Informed Graph Neural Network Framework for Real- Time Thermal-Aware Fault Prediction and Adaptive Power Optimization in Heterogeneous System-on-Chip Architectures. Journal of VLSI Design Tools and Technology. 2026; 16(02):-. Available from: https://journals.stmjournals.com/jovdtt/article=2026/view=250870

References

  1. Antolini A, Lico A, Zavalloni F, Scarselli EF, Gnudi A, Torres ML, Canegallo R, Pasotti M. A readout scheme for PCM-based analog in-memory computing with drift compensation through reference conductance tracking. IEEE Open Journal of the Solid-State Circuits Society. 2024 Jul 25;4:69-82.
  2. Pistolesi L, Ravelli L, Glukhov A, de Gracia Herranz A, Lopez-Vallejo M, Carissimi M, Pasotti M, Rolandi PL, Redaelli A, Martín IM, Bianchi S. Differential phase change memory (PCM) cell for drift- compensated in-memory computing. IEEE Transactions on Electron Devices. 2024 Oct 25;71(12):7447-53.
  3. Yang X, Lei Y, Yu Q, Wang Q, Chen H, Song Z. Phase change memory programming circuit with improved speed. Moore and More. 2025 Jan 15;2(1):3.
  4. Syed GS, Le Gallo M, Sebastian A. Phase-change memory for in-memory computing. Chemical reviews. 2025 May 22;125(11):5163-94.
  5. Zhou W, Shen X, Yang X, Wang J, Zhang W. Fabrication and integration of photonic devices for phase- change memory and neuromorphic computing. International Journal of Extreme Manufacturing. 2024 Apr 1;6(2):022001.
  6. Zhou W, Shen X, Yang X, Wang J, Zhang W. Fabrication and integration of photonic devices for phase- change memory and neuromorphic computing. International Journal of Extreme Manufacturing. 2024 Apr 1;6(2):022001.
  7. Le Gallo M, Khaddam-Aljameh R, Stanisavljevic M, Vasilopoulos A, Kersting B, Dazzi M, Karunaratne G, Brändli M, Singh A, Mueller SM, Büchel J. A 64-core mixed-signal in-memory compute chip based on phase-change memory for deep neural network inference. Nature Electronics. 2023 Sep;6(9):680-93.
  8. Wu C, Deng H, Huang YS, Yu H, Takeuchi I, Ríos Ocampo CA, Li M. Freeform direct-write and rewritable photonic integrated circuits in phase-change thin films. Science Advances. 2024 Jan 5;10(1):eadk1361.
  9. Miller F, Chen R, Fröch JE, Rarick H, Geiger S, Majumdar A. Rewritable photonic integrated circuits using dielectric-assisted phase-change material waveguides. Optics Letters. 2023 Apr 26;48(9):2385-8.10.
  10. Wu C, Yu H, Lee S, Peng R, Takeuchi I, Li M. Programmable phase-change metasurfaces on waveguides for multimode photonic convolutional neural network. Nature communications. 2021 Jan 4;12(1):96.
  11. Xie H, Wang Y, Gao Z, Ganthia BP, Truong CV. Research on frequency parameter detection of frequency shifted track circuit based on nonlinear algorithm. Nonlinear Engineering. 2021 Jan 1;10(1):592-9..
  12. Gu J, Wang W, Yin R, Truong CV, Ganthia BP. Complex circuit simulation and nonlinear characteristics analysis of GaN power switching device. Nonlinear Engineering. 2021 Jan 1;10(1):555- 62..
  13. Rubavathy SJ, Venkatasubramanian R, Kumar MM, Ganthia BP, Kumar JS, Hemachandu P, Ramkumar MS. Smart grid based multiagent system in transmission sector. In2021 Third International Conference on Inventive Research in Computing Applications (ICIRCA) 2021 Sep 2 (pp. 1-5). IEEE.
  14. Zheng W, Mehbodniya A, Neware R, Wawale SG, Ganthia BP, Shabaz M. Modular unmanned aerial vehicle platform design: Multi-objective evolutionary system method. Computers and Electrical Engineering. 2022 Apr 1;99:107838.
  15. Ranjan S, Jaiswal S, Latif A, Das DC, Sinha N, Hussain SS, Ustun TS. Isolated and interconnected multi-area hybrid power systems: A review on control strategies. Energies. 2021 Dec 8;14(24):8276.
  16. Pellis S. Golden Fractals in Fluid Dynamics and Turbulence. Available at SSRN 5543579. 2025 Sep 28.
  17. Sharma BP, Peelam MS, Gupta A, Shekhar C, Chamola V. A comprehensive survey on data converters for iot applications: Scope, issues and future directions. IEEE Internet of Things Journal. 2025 Mar 20.
  18. Baraa SM, Desa H, Mohammed KS, Al-Malaisi TA, Hussain AS, Majdi HS. Selective harmonic elimination in reduced-switch multilevel inverters for PV systems using the sparrow search algorithm. Journal of Robotics and Control (JRC). 2025 Feb 15;6(1):385-95.
  19. Elahi M, Trinh-Van S, Yang Y, Lee KY, Hwang KC. Compact and high gain 4× 4 circularly polarized microstrip patch antenna array for next generation small satellite. Applied Sciences. 2021 Sep 23;11(19):8869.
  20. Wu J, Liang Y, Liu G, Pang R, Teng Y, Li C, Bao X, Lei S, Cai Z. Research on Intelligent Thermal Optimization for Chiplet-Based Heterogeneously Integrated AI Chip Embedded with Leaf-Vein- Inspired Fractal Microchannels. Materials. 2026 Feb 10;19(4):679.
  21. Zhang H, Ge Y, Zhao X, Wang J. Hierarchical deep reinforcement learning for multi-objective integrated circuit physical layout optimization with congestion-aware reward shaping. IEEE Access. 2025 Sep 16.
  22. Incorvia JA, Xiao TP, Zogbi N, Naeemi A, Adelmann C, Catthoor F, Tahoori M, Casanova F, Becherer M, Prenat G, Couet S. Spintronics for achieving system-level energy-efficient logic. Nature Reviews Electrical Engineering. 2024 Nov;1(11):700-13.
  23. Liu W, Liu Z, Lv S. Thermal monitoring modeling of solid-state batteries: Decoding temperature inhomogeneity via machine learning of interfacial heat generation. Applied Energy. 2026 Jan 15;403:127037.
Ahead of Print Subscription Review Article
Volume 16
02
Received 15/07/2026
Accepted 16/07/2026
Published 28/07/2026
Publication Time 13 Days

Login

My IP

PlumX Metrics