DocumentCode :
228602
Title :
Central Switch Noded Mesh architecture (CSNM)
Author :
Nambiar, Sandeep Gopi ; Swaminathan, K. ; Lakshminarayanan, G. ; Seok-Bum Ko
Author_Institution :
Dept. of ECE, Nat. Inst. of Technol., Tiruchirappalli, India
fYear :
2014
fDate :
13-14 Feb. 2014
Firstpage :
1
Lastpage :
5
Abstract :
In this paper, a novel mesh based architecture named as Central Switch Noded Mesh architecture (CSNM) with Switching Nodes (SWN) placed at the center of each square block of mesh topology is proposed. The SWNs form a subnetwork inside the actual mesh network and examines the global traffic and informs the router as it routes its packets adaptively. We have also defined an adaptive routing algorithm for unicasting(UC) and a multicasting (MC) routing. The architecture along with its adaptive routing algorithm detects deadlock and avoids it by deviating the traffic through the diagonal link. The proposed routing algorithm is fully adaptive as the packets are free to move in all the directions (except the incoming directions). MC routing algorithm lessens latency and improves performance when compared to other MC based routing algorithms. The result analysis shows an improvement in throughput and reduction of latency with anegligible area overhead.
Keywords :
network routing; network topology; network-on-chip; CSNM; MC routing; UC routing; adaptive routing algorithm; central switch noded mesh architecture; deadlock detection; diagonal link; latency reduction; mesh based architecture; mesh network; mesh topology; multicasting routing; network on chip; router; unicasting routing; Adaptation models; Artificial neural networks; Computer architecture; Ports (Computers); Routing; Switches; Adaptive; Congestion; Global traffic; Multicasting; Network on Chip; Switching Nodes;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Electronics and Communication Systems (ICECS), 2014 International Conference on
Conference_Location :
Coimbatore
Print_ISBN :
978-1-4799-2321-2
Type :
conf
DOI :
10.1109/ECS.2014.6892693
Filename :
6892693
Link To Document :
بازگشت