Distributed Slot Synchronization:
A Network-Wide Synchronization Technique for Packet-Switched Optical Networks
Tad Hofmeister
Department of Electrical Engineering
Abstract
The phenomenal growth in data networking traffic over the last five years has left current telecommunication networks struggling to keep up with
demand. While optical Wavelength Division Multiplexing (WDM) technology is now being deployed in backbone trunks to exploit the enormous
bandwidth of optical fiber, all switching and routing functions are still performed electronically, limiting network capacity. In order for networks to
meet future traffic loads, some of the routing and switching functions must be performed in the optical domain. WDM packet-switched networks
offer a potential solution to bypass the electronics bottleneck, providing orders of magnitude more switching capacity.
One of the key challenges facing the implementation of WDM packet-switched networks is efficient synchronization of packets. In order to be
switched, packets must arrive at switch inputs aligned to one another in time. In electronic switches, variable-length delay buffers are used at the
switch inputs to synchronize arriving packets. However, equivalent optical variable-length delay buffers are not yet practically realizable.
Distributed Slot Synchronization (DSS) is a novel alternative approach that does not require variable-length optical buffers. Instead, all nodes
coordinate their transmission times so that packets arrive aligned to one another, independent of the distances packets travel and the
propagation delays they encounter.
The DSS technique is robust and is data rate and format independent. DSS was developed for use in a 2.5 Gb/s/wavelength, WDM optical
packet-switched network. In this implementation, it was experimentally demonstrated that the DSS system, operating with 80 MHz control logic,
achieves a packet arrival jitter of less than 13 ns with 12 km node spacings. In this talk, the design, construction, investigation, and results of the
DSS system will be presented. Performance optimization and applications of DSS in other network architectures will also be discussed.
|