PentaNoc: A New Scalable and self-similar NoC Architecture

Ahlem Boudellioua, Nasser Alzeidi*

*Corresponding author for this work

Research output: Contribution to journalConference articlepeer-review

3 Citations (Scopus)


In massively multi-core System-on-Chip (SoCs), high throughput, low delay and fault tolerance are essential factors to fully harness the computational power offered by tens and hundreds of processing elements (PEs). Network-on-Chip (NoCs) have emerged as a recognized solution to fulfill the ever growing communication demands to interconnect the PEs in modern SoCs. In this research we proposed and analytically evaluated a novel self-similar and highly scalable NoC topology (called PentaNoC) with desirable topological properties including, low node degree, average hop count, diameter and high path diversity as well as bisection width. The new topology is based on cascading any number of pentagon-shape blocks. The research involved formulating the topological properties of the proposed topology, devising a deadlock-free routing protocol and comparing the performance of the proposed topology with its counterparts. Against three topologies, the PentaNoC topology showed better results in terms of having the smallest average hop count and diameter, and the maximum bisection width. The analytical evaluation has also shown that the zero-load latency and the ideal throughput of the proposed architecture outperforms its counterparts.

Original languageEnglish
Pages (from-to)358-364
Number of pages7
JournalProcedia Computer Science
Publication statusPublished - 2018
Event15th International Conference on Mobile Systems and Pervasive Computing, MobiSPC 2018 - Las Palmas de Gran Canaria, Spain
Duration: Aug 13 2018Aug 15 2018


  • Analytical Modelling
  • Network Topology
  • Networks on Chip
  • Performance Evaluation

ASJC Scopus subject areas

  • General Computer Science


Dive into the research topics of 'PentaNoc: A New Scalable and self-similar NoC Architecture'. Together they form a unique fingerprint.

Cite this