Optical Packet Switching Technology for Future Global Sensing Networks

The amount of Internet traffic in current backbone networks has grown enormously due to wide deployment of broadband Internet access and new multimedia applications such as sensing data sharing and real-time streaming. In regard to optical transport technology based on Wavelength Division Multiplex (WDM), available bandwidth within single fiber could be over 20Tb/s, according to the most recent reports [1, 2]. Consequently, the router which handles such high-capacity traffic in current and future networks is necessary. Optical packet switching (OPS) technology has been researched in order to efficiently utilize such broad bandwidth with lower power consumption than current switching architecture with O/E/O conversion and higher link-utilization ratio than optical circuit/burst switching (OCS/OBS) . In a network based on general OPS architectures, an optical packet consisting of a header and a payload is encoded into single wavelength. An independent optical packet is, therefore, encoded into each wavelength available within single fiber. This property makes several kinds of devices necessary for function an incoming packet such as header processing and contention resolution proportional to W ×P , where W is the number of the wavelengths available in the WDM network and P is the number of input ports in the core node. The increase of the number of these devices could make the implementation of the node unrealistic in terms of physical size and cost especially in the future WDM networks in which the number of available wavelengths in a single fiber is over a thoughsand[3]. Multi-Wavelength OPS (MW-OPS) has been researched as one of solustions that can reduce the number of these devices. Core nodes based on MW-OPS use optical switchies independent of wavelengths and forward optical packet encoded into multiple wavelengths, for instance, header data is encoded into a wavelength and payload data is encoded into other wavelengths. This characteristics makes the necessary number of optical devices processing incoming packets in a core node propotional to P , compared to traditional OPS architecture. In addition to that, the control complexity in MW-OPS nodes could be lower than that in traditional OPS nodes. Due to these characteristics, MW-OPS could be considered as more scalable architecture. There are currently three proposed node architectures based on MW-OPS: Onaka’s 100 (10λ× 10) Gb/s MW-OPS node using SOA switches[4], Furukawa’s 160 (16λ× 10) Gb/s MW-OPS node using LiNbO3 switches[5], and our 80 (8λ× 10) Gb/s MW-OPS node using PLZT switches[6]. Each of them has comfirmed the feasibility of MW-OPS using the correspondng type of switches. In this paper, we demonstrate 320 (8λ×40) Gb/s MW-OPS with contention resolution mechanism using PLZT switches in order to evaluate its feasibility. The major characteristics of PLZT switches are four: 1) high switching speed (< 6ns), 2) low polarization dependency, 3) noise robustness, and 4) low drive voltage. On the other hand, SOA switches have the issue on OSNR degradation due to ASE noise and LN switches require higher drive voltage for removing polarization dependency, while both provide higher switching speed (< 1ns). In addtion, DC drift phenomenon can not be ignored for LN switches. Therefore, PLZT switches are considered practical and suitable for OPS[7].