Ultra-low latency optical packet switching node

Abstract

Abstract-A photonic packet switching node is introduced, and its routing latency is shown to be 15.3 ns. The power penalty of the node at a bit-error rate (BER) of 10 9 is confirmed to be approximately 0.2 dB across 33 nm of the -band for 10-Gb/s payload wavelengths. Moreover, multiple-wavelength packets containing 16 payload wavelengths can be switched while maintaining BERs of 10 12 or better.Index Terms-Interconnection networks, packet switching, photonic switching systems, wavelength-division multiplexing (WDM).

A Modular, Scalable, Extensible, and Transparent Optical Packet Buffer

“…From this selection of routing options, it can be deduced that each SOA hop introduces approximately 0.4 dB of power penalty on average. This result is similar to those for other simplistic SOA-based OPS nodes and modules [11], [15], [17], [23], [24]. Also, in order to demonstrate the wideband transparency of the buffer implementation, eye diagrams for different payload wavelengths are presented as well (Fig.…”

“…These packets can also have a multiple-wavelength (wavelength-striped) format, as is found in [15], [16], and [22].…”

“…Because these losses depend on which branches are required for a particular module implementation, the SOAs are set to deliver between 6 and 11 dB of gain, which requires between 45 and 75 mA of drive current. The net gain or loss incurred on each packet payload can be kept to less than about 0.5 dB, as in [15], [16], and [24].…”

“…The individual modules execute routing logic based only on the incoming read request signal and the presence or absence of packets on each of the three inputs (including the internal buffer); the routing logic itself is memoryless (i.e., combinatorial logic). This kind of modular and distributed structure leverages the same paradigm as [15]- [18]. Table I details the truth table needed to implement the routing logic in each module as required for the queue (FIFO) functionality; Table II implements a stack (LIFO).…”

The Data Vortex, an All Optical Path Multicomputer Interconnection Network

“…The optical components are capable of working over a bandwidth of about 6 THz near 1,550 nm (the ITU C-band), allowing for packets with extremely large data rates (i.e., terabits per second [12]) to be transmitted simultaneously through the same components and optical fiber. Packets are composed of many independently modulated channels (wavelengths) transmitted in parallel; particular channels are used for routing headers, and others are designated for the payload [26], [27].…”