WDM optical packet ring performance insights: Scheduling and capacity

• Published in: 2012 IEEE Symposium on Computers and Communications (ISCC) pp. 000253 - 000258
• Main Authors: Uscumlic, B., Matavulj, P., Gravey, A., Gravey, P., Morvan, M.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-07-2012
• Subjects: Computer Science; Delay; metropolitan rings; Networking and Internet Architecture; Optical buffering; Optical packet switching; Optical receivers; optical transparency; Round robin; Scheduling; Wavelength division multiplexing; WDM; WDM; Metropolitan rings; Scheduling; Metropolitan area network; Optical packet switching; Optical transparency
• ISBN: 9781467327121; 1467327123
• ISSN: 1530-1346; 2642-7389
• DOI: 10.1109/ISCC.2012.6249304

This paper addresses WDM optical packet rings which present a single control channel and multiple data channels. These rings are intended as transport networks in metro areas, i.e. are considered as replacement for the traditional SONET/SDH or Ethernet rings. These networks are thus expected to deliver a lossless and low latency service. In the considered transfer mode, time is slotted, traffic insertion is opportunistic and transit traffic always has priority over new traffic which ensures optical transparency. The present work assesses the impact on delay performance of the two configuration characteristics of these rings: scheduling mechanism and number of data channels. We compare two Max-Weight scheduling policies (Oldest Packet First, OPF and the originally proposed Largest Virtual Waiting Time First, LVWTF) with the commonly used, non Max-Weight, Round-Robin (RR) scheduling in order to assess the respective Fairness performances of LVWTF and RR. It is shown that unless the ring is heavily loaded, RR is as fair, or fairer than LVWTF. In the second part of the paper, we have analyzed the performance of an optical packet ring with uniform and complete traffic matrix configured with LVWTF scheduling. The mean delay performance is quantified, and shown to depend both on the number of data channels and on the number of nodes in the ring. In particular, it is shown that for a given number nodes, and a given load, the delay presents a minimum for a finite number of data channels, which shows that increasing resources (number of data channels) may indeed degrade QoS (delay).