SRCPAR-Spike Response-Based Congestion Prediction for Adaptive Routing for 2D NoCs
Saved in:
| Title: | SRCPAR-Spike Response-Based Congestion Prediction for Adaptive Routing for 2D NoCs |
|---|---|
| Authors: | Singh, Rajendra, Bohra, Manoj Kumar, Sharma, Ashish, Verma, Sourabh Singh, Bhatt, Devershi Pallavi, Bairwa, Amit Kumar, Daneshtalab, Masoud |
| Source: | IEEE Access. 13:114542-114554 |
| Subject Terms: | Routing, Prediction algorithms, Traffic control, Real-time systems, Clocks, Throughput, Multicast algorithms, Load modeling, Heuristic algorithms, Hardware, NoC, congestion prediction, ANN, industry innovation |
| Description: | Network-on-Chip (NoC) architectures offer several advantages over bus-based systems, including improved scalability, efficient and high-performance interconnects, reduced wire routing congestion, and enhanced power efficiency. These benefits make NoC a promising solution for complex System-on-Chip (SoC) designs. In NoC, congestion prediction and its use in routing algorithms play a significant role in determining performance. A major challenge is the staleness of congestion information when it reaches a router. This paper introduces a Spike Response-based Congestion Prediction and Adaptive Routing (SRCPAR) algorithm to mitigate this issue. It leverages spike information alongside the local congestion data to accurately predict network congestion. Performance is evaluated using metrics such as average latency and throughput. Proposed method shows an improvement in performance for synthetic and real traffic patterns. |
| File Description: | |
| Access URL: | https://urn.kb.se/resolve?urn=urn:nbn:se:mdh:diva-73105 https://doi.org/10.1109/ACCESS.2025.3583432 |
| Database: | SwePub |
Full Text Finder 
Nájsť tento článok vo Web of Science | Abstract: | Network-on-Chip (NoC) architectures offer several advantages over bus-based systems, including improved scalability, efficient and high-performance interconnects, reduced wire routing congestion, and enhanced power efficiency. These benefits make NoC a promising solution for complex System-on-Chip (SoC) designs. In NoC, congestion prediction and its use in routing algorithms play a significant role in determining performance. A major challenge is the staleness of congestion information when it reaches a router. This paper introduces a Spike Response-based Congestion Prediction and Adaptive Routing (SRCPAR) algorithm to mitigate this issue. It leverages spike information alongside the local congestion data to accurately predict network congestion. Performance is evaluated using metrics such as average latency and throughput. Proposed method shows an improvement in performance for synthetic and real traffic patterns. |
|---|---|
| ISSN: | 21693536 |
| DOI: | 10.1109/ACCESS.2025.3583432 |