EUVIMEDEUVIMED entdecken
Crossref

A Multimodal Fusion Graph Convolutional Network With Pufferfish Optimization Algorithm–Based Latency Aware Predictive Routing Model for Network‐on‐Chip Architectures

P. Sivagamasundhari, M. Santhi

International Journal of Communication Systems · 10.09.2026 · 39 (13)

DOI: 10.1002/dac.70566 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Communication Systems, online erschienen am 29.07.2026. Verfasst von P. Sivagamasundhari und 1 weiteren Autor:innen. Die Arbeit nennt 33 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Network‐on‐chip (NoC) architectures are the building blocks of high‐performance computing systems, but routing inefficiencies tend to increase latency and congestion, which ultimately constrain overall performance. In this manuscript, a novel enhanced deep learning‐based latency‐aware predictive routing model (LAPR) is presented that adapts packet flow dynamically to minimize communication delays in NoC architectures. The developed architecture exploits a multimodal fusion graph convolutional network (MF‐GCN) to model source and channel sequence feature vectors for the prediction of NoC traffic patterns. MF‐GCN captures spatial and sequential details through features associated with source and channel queuing delays and significantly enhances prediction accuracy. In addition, effective routing routes are dynamically optimized through the utilization of the presented PufferFish Optimization Algorithm (POA) with exploration and exploitation trade‐offs according to varying network conditions, thus improving network performance. Consequently, the optimized LAPR greatly enhances communication efficiency in NoC structures. The model is implemented in Python, and the performance metrics of latency, throughput, squared correlation coefficient (SCC), and normalized root mean square error (NRMSE) are measured. The proposed MF‐GCN‐POA recorded 22.72% less energy consumption, 18.63% less NRMSE, 27.23% higher SCC, 27.09% less latency, and 23.77% better throughput. All these outcomes demonstrate the model's capability to reduce communication efficiency with dynamic traffic conditions in mind, indicating its potential as a next‐generation high‐performance computing system solution and for intricate NoC environments.

Zitieren

APA

Sivagamasundhari, P., & Santhi, M. (2026). A Multimodal Fusion Graph Convolutional Network With Pufferfish Optimization Algorithm–Based Latency Aware Predictive Routing Model for Network‐on‐Chip Architectures. International Journal of Communication Systems, 39(13). https://doi.org/10.1002/dac.70566

Vancouver

Sivagamasundhari P, Santhi M. A Multimodal Fusion Graph Convolutional Network With Pufferfish Optimization Algorithm–Based Latency Aware Predictive Routing Model for Network‐on‐Chip Architectures. International Journal of Communication Systems. 2026;39(13). doi:10.1002/dac.70566

BibTeX
@article{Sivagamasundhari2026,
  author = {Sivagamasundhari, P. and Santhi, M.},
  title = {A Multimodal Fusion Graph Convolutional Network With Pufferfish Optimization Algorithm–Based Latency Aware Predictive Routing Model for Network‐on‐Chip Architectures},
  journal = {International Journal of Communication Systems},
  year = {2026},
  volume = {39},
  number = {13},
  doi = {10.1002/dac.70566}
}
Kostenlos starten →