Design and Performance of Randomized Network Schedules for Time-domain Wavelength Interleaved Networks

Time-domain Wavelength Interleaved Network (TWIN) is a newly proposed all-optical network architecture that provides transport services for synchronous (fixed-rate) and asynchronous (bursty) traffic. Each node has a tunable laser to transmit bursts to various destinations and a burst-mode receiver capable of decoding each burst intended to it. The interior nodes route each burst independently and passively based on the wavelength of the burst. Network scheduling in TWIN is a critical component that arbitrates burst transmissions from sources to destinations so that conflicts are either avoided or minimized and resource efficiency is maximized. In this paper, we propose a distributed protocol that performs scheduling of bursts in a network with non-zero propagation delays. The protocol uses a combination of randomization and prioritization techniques to allow for reassignments of time slots. In one realization, each node randomly assigns non-overlapping time slots for its transmissions. In another realization, each destination grants random but non-overlapping time slots to sources for transmissions. In neither scheme is there any coordination across sources and destinations. We develop an analytical formulation of these schemes based on a new occupancy model and investigate the performance of the proposed distributed protocols under two broad types of traffic: fixed load traffic and bursty traffic. The analytical results show that there is 67% efficiency for the distination-based and 33% efficiency for the source-based protocols for bursty and unpredictable traffic. When end-to- end demands are nearly static, we show through simulations that the destination-based protocol can converge to nearly 100% efficiency via learning. These results indicate that the proposed distributed protocols, though exceedingly simple, provide acceptable efficiency comparable to that realized by (computation-intensive) centralized schedulers irrespective of propagation delays and applicable to any kind of network traffic.