Throughput comparison of Shuffle-exchange networks with additional stages due to resource scheduling

Abstract

Several approaches are introduced to increase the throughput and reliability of Multistage Interconnection Networks (MINs), mostly by changing the network architecture. When multiple sources in such a network try to send data, collision of packets and blocking problems are inevitable. Time division multiple access (TDMA) protocol are used to address these issues. However, TDMA protocol based on fixed slot allocation has low throughput and high delay in computer networks with high load. To improve the performance of this protocol, we propose an adaptive slot allocation strategy using Monte Carlo random sampling method. We apply the proposed slot allocation approach on a well-known type of MIN, Shuffle-exchange network (SEN). Then, the effects of proposed approach on SEN with one additional stage (SEN+), and SEN with two additional stages (SEN+2) are investigated. Simulation results show that the SEN+(TDMA) and SEN+2(TDMA) outperform the regular ones in terms of throughput and reliability. SEN+2(TDMA) also shows improvement in network throughput and reliability compared to regular SEN+2, about 3% and 2% respectively.