Back to Search Start Over

PCOI: Packet Classification‐Based Optical Interconnect for Data Centre Networks.

Authors :
Nawaz Jadoon, Rab
Fayyaz, Mohsin
Zhou, WuYang
Khan, Muhammad Amir
Mujtaba, Ghulam
Source :
Mathematical Problems in Engineering. 7/17/2020, p1-11. 11p.
Publication Year :
2020

Abstract

To support cloud services, Data Centre Networks (DCNs) are constructed to have many servers and network devices, thus increasing the routing complexity and energy consumption of the DCN. The introduction of optical technology in DCNs gives several benefits related to routing control and energy efficiency. This paper presents a novel Packet Classification based Optical interconnect (PCOI) architecture for DCN which simplifies the routing process by classifying the packet at the sender rack and reduces energy consumption by utilizing the passive optical components. This architecture brings some key benefits to optical interconnects in DCNs which include (i) routing simplicity, (ii) reduced energy consumption, (iii) scalability to large port count, (iv) packet loss avoidance, and (v) all-to-one communication support. The packets are classified based on destination rack and are arranged in the input queues. This paper presents the input and output queuing analysis of the PCOI architecture in terms of mathematical analysis, the TCP simulation in NS2, and the physical layer analysis by conducting simulation in OptiSystem. The packet loss in the PCOI has been avoided by adopting the input and output queuing model. The output queue of PCOI architecture represents an M/D/32 queue. The simulation results show that PCOI achieved a significant improvement in terms of throughput and low end-to-end delay. The eye-diagram results show that a good quality optical signal is received at the output, showing a very low Bit Error Rate (BER). [ABSTRACT FROM AUTHOR]

Details

Language :
English
ISSN :
1024123X
Database :
Academic Search Index
Journal :
Mathematical Problems in Engineering
Publication Type :
Academic Journal
Accession number :
144613717
Full Text :
https://doi.org/10.1155/2020/2903157