Traffic in data centers networks (DCNs) is steadily growing to support various applications and virtualization technologies. Multi-tenancy enabling efficient resource utilization is considered as a key requirement for the next generation DCs resulting from the growing demands for services and applications. Virtualization mechanisms and technologies can leverage statistical multiplexing and fast switch reconfiguration to further extend the DC efficiency and agility. We present a novel high performance flat DCN employing bufferless and distributed fast (sub-microsecond) optical switches with wavelength, space, and time switching operation. The fast optical switches can enhance the performance of the DCNs by providing large-capacity switching capability and efficiently sharing the data plane resources by exploiting statistical multiplexing. Benefiting from the Software-Defined Networking (SDN) control of the optical switches, virtual DCNs can be flexibly created and reconfigured by the DCN provider. Numerical and experimental investigations of the DCN based on the fast optical switches show the successful setup of virtual network slices for intra-data center interconnections. Experimental results to assess the DCN performance in terms of latency and packet loss show less than 10^-5 packet loss and 640ns end-to-end latency with 0.4 load and 16- packet size buffer. Numerical investigation on the performance of the systems when the port number of the optical switch is scaled to 32x32 system indicate that more than 1000 ToRs each with Terabit/s interface can be interconnected providing a Petabit/s capacity. The roadmap to photonic integration of large port optical switches will be also presented.
Based on Amdahl scaling of tree-networks, we show that in the next 10 years power efficiency and cost of data center communication networks have to improve with three orders of magnitude. Flattened network architectures may allow for more efficient scaling but require high-radix network switches. In turn, such switches will require opto-electronic conversion in close proximity of the switch ASIC. In this paper, we focus on three-dimensional die-stacked transceiver ICs that allow for low cost fabrication and packaging that may enable flattened network architectures based on highradix switches.
Based on well-known laws of physics, a lower bound on the energy-per-bit required for
transmitting information using a photonic channel is established. The analysis includes the
energy required to convert information from the electronic to the photonic domain and back.
We investigate links that employ a directly modulated laser as well as links that employ an
external modulator. It is shown that the power dissipation of the channel also imposes a
bound on the maximum bandwidth density for a photonic link. Keeping this in mind,
opportunities for optics in computing systems are discussed, especially from a systems
perspective.
We demonstrate a new approach to increase the optical interconnection bandwidth density by stacking the opto-electrical
dies directly on the CMOS driver. The suggested implementation is aiming to provide a wafer scale process which will
make the use of wire bonding redundant and will allow for impedance matched metallic wiring between the electronic
driving circuit and its opto-electronic counter part. We suggest the use of a thick photoresist ramp between CMOS driver
and opto-electrical dies surface as the bridge for supporting co-plannar waveguides (CPW) electrically plated with
lithographic accuracy. In this way all three dimensions of the interconnecting metal layer, width, length and thickness
can be completely controlled. In this 1st demonstration all processing is done on commercially available devices and
products, and is compatible with CMOS processing technology.
To test the applicability of CPW instead of wire bonds for interconnecting the CMOS circuit and opto-electronic chips,
we have made test samples and tested their performance at speeds up to 10 Gbps. In this demonstration, a silicon
substrate was used on which we evaporated gold co-planar waveguides (CPW) to mimic a wire on the driver. An optical
link consisting of a VCSEL chip and a photodiode chip has been assembled and fully characterized using optical
coupling into and out of a multimode fiber (MMF). A 10 Gb/s 27-1 NRZ PRBS signal transmitted from one chip to
another chip was detected error free. A 4 dB receiver sensitivity penalty is measured for the integrated device compared
to a commercial link.
HISTORIC aims to develop and test complex photonic integrated circuits containing a relatively large number
of digital photonic elements for use in e.g. all-optical packet switching. These photonic digital units are alloptical
flip-flops based on ultra compact laser diodes, such as microdisk lasers and photonic crystal lasers.
These lasers are fabricated making use of the heterogeneous integration of InP membranes on top of silicon
on insulator (SOI) passive optical circuits. The very small dimensions of the lasers are, at least for some
approaches, possible because of the high index contrast of the InP membranes and by making use of the
extreme accuracy of CMOS processing.
All-optical flip-flops based on heterogeneously integrated microdisk lasers with diameter of 7.5μm have
already been demonstrated. They operate with a CW power consumption of a few mW and can switch in 60ps
with switching energies as low as 1.8 fJ. Their operation as all-optical gate has also been demonstrated.
Work is also on-going to fabricate heterogeneously integrated photonic crystal lasers and all-optical flip-flops
based on such lasers. A lot of attention is given to the electrical pumping of the membrane InP-based photonic
crystal lasers and to the coupling to SOI wire waveguides. Optically pumped photonic crystal lasers coupled
to SOI wires have been demonstrated already.
The all-optical flip-flops and gates will be combined into more complex photonic integrated circuits,
implementing all-optical shift registers, D flip-flops, and other all-optical switching building blocks.
The possibility to integrate a large number of photonic digital units together, but also to integrate them with
compact passive optical routers such as AWGs, opens new perspectives for the design of integrated optical
processors or optical buffers. The project therefore also focuses on designing new architectures for such
optical processing or buffer chips.
The European BOOM project aims at the realization of high-capacity photonic routers using the silicon material as the
base for functional and cost-effective integration. Here we present the design, fabrication and testing of the first BOOMgeneration
of hybrid integrated silicon photonic devices that implement key photonic routing functionalities. Ultra-fast
all-optical wavelength converters and micro-ring resonator UDWDM label photodetectors are realized using either 4um
SOI rib or SOI nanowire boards. For the realization of these devices, flip-chip compatible non-linear SOAs and
evanescent PIN detectors have been designed and fabricated. These active components are integrated on the SOI boards
using high precision flip-chip mounting and heterogeneous InP-to-silicon integration techniques. This type of scalable
and cost-effective silicon-based component fabrication opens up the possibility for the realization of chip-scale, power
efficient, Tb/s capacity photonic routers.
We present a CLOS architecture for optical packet switching that allows for large throughput, low latency and contention
resolution in the optical domain. The switch is scalable in terms of bit-rate and data-format and port-number. We
describe devices and subsystems that support this architecture.
We present the results of a transmission experiment, over 110 km of field installed fiber, for an all-optical 160 Gb/s
packet switching system. The system uses in-band optical labels which are processed entirely in the optical domain
using a narrow-band all-optical filter. The label decision information is stored by an optical flip-flop, which output
controls a high-speed wavelength converter based on ultra-fast cross-phase modulation in a single semiconductor optical
amplifier. The packet switched node is located in between two different fiber sections, each having a length of 54.3-km.
The field installed fibers are located around the city of Eindhoven in the Netherlands. The results show how the all-optical
switch can effectively route the packets based on the optical information and that such packets may be
transmitted across the fiber with an acceptable penalty level.
The routing decision functionality by all-optically interconnecting semiconductor-based all-optical logic gates and flip-flops is demonstrated in the frame of an all-optical packet switching network. We experimentally show that the output of the all-optical 2-bit correlator is capable of toggling the states of the integrated flip-flop every 2.5 ns via an adaptation stage. High extinction ratios are obtained at the output of the flip-flop, which can be used to feed a high-speed wavelength converter to complete the routing functionality of the AOLS node. The potential integration of these SOA-MZI based devices make the proposed approach a very interesting solution for future packet switched optical networks.
We present some progress in the field of optical signal processing that could be utilized in all-optical packet switching. We demonstrate error-free 160 Gb/s optical wavelength conversion employing a single semiconductor optical amplifier. The gain recovery time of the semiconductor optical amplifier is greater than 90 ps. Assisted by an optical bandpass filter, an effective recovery time of 3 ps is achieved in the wavelength converter, which ensures 160 Gb/s operation. This optical wavelength converter can be controlled by a monolithically integrated optical flip-flop memory to route 80 Gb/s data-packets all-optically. The routing is realized without electronic control. The integrated optical flip-flop is based on two-coupled lasers, exhibits single-mode operation, has 35 dB contrast ratio between the states and switches state in about 2 ns. We demonstrate that the integrated flip-flop is able to control the optical wavelength converter up to 160 Gb/s. The system is capable of routing 80 Gb/s data packets with duration of 35 ns, separated by 15 ns of guard time.
We discuss how all-optical signal processing might play a role in future all-optical packet switched networks. We introduce a concept of optical packet switches that employ entirely all-optical signal processing technology. The optical packet switch is made out of three functional blocks: the optical header processing block, the optical memory block and the wavelength conversion block. The operation principle of the optical packet switch is explained. We show that these three functional blocks can be realized by using the nonlinearities of semiconductor optical amplifiers. Some technologies in these three functional blocks are described. The header processor is realized using a Terahertz Optical Asymmetric Demultiplexer. We also describe a header pre-processor to improve the extinction ratio of the header processor output. In the optical memory block, we show that an all-optical memory can be obtained by using two coupled lasers that form a master-slave configuration. The state of the optical memory is distinguished by the wavelength of the master laser. We extend the concept to an optical memory can have multiple states. In the wavelength conversion block, we demonstrate a 160 Gbit/s wavelength conversion using a single semiconductor optical amplifier in combination with a well-designed optical bandpass filter. The semiconductor optical amplifier has a gain recovery time
greater than 90 ps, which corresponds to a less than 20 GHz bandwidth for conventional wavelength conversion. We show that by properly using the optical bandpass filter, ultrafast dynamics in the semiconductor optical amplifier can be employed for wavelength conversion at ultrahigh bit-rates.
We discuss how all-optical signal processing might play a role in future all-optical packet switched networks. We describe a few approaches to optical header processing, all based on nonlinearities in a semiconductor optical amplifier. In first approach a SLALOM configuration is used. The second approach uses a Terahertz Optical Asymmetric Demultiplexer. We also describe a header pre-processor to improve the extinction ratio of the header processor output. The second functional block on which we focus is optical buffering. We show how all-optical signal processing technology can be used to route a packet into a fiber delay line and we describe a circulating optical loop based op optical technology.
We explain how a semiconductor optical amplifier in a Sagnac-interferometric arrangement can be used for switching of 200 fs optical pulses. The switching principles are based on gain and index saturation dynamics on a sub-picosecond timescale. We present a model that accounts for bi-directional propagation of ultrashort optical pulses through the amplifier as well as free-carrier absorption and two-photon absorption. We have also carried out pump and probe experiments to measure the ultrafast refractive index dynamics of a multi-quantum well InGaAsP-InGaAs semiconductor optical amplifier that is operated in the gain regime. The pump and probe pulses are cross-linearly polarized. We observe a phase shift of 200 degrees if the amplifier is pumped with 120 mA of current, but find that the phase shift vanishes if the injection current is increased to 160 mA. Our results indicate a contribution of two-photon absorption to the nonlinear phase shift that opposes the phase shift introduced by the gain. Finally, we observe that the phase shift comes up and disappears within a picosecond.
Telecommunication systems using 200 fs optical pulses for ultrahigh bit-rate optical transmission require new concepts to be developed for ultrafast all-optical switching. On the basis of numerical experiments, we discuss how a semiconductor optical amplifier in a Sagnac interferometric arrangement can be used for switching these short pulses. The switching principle is based on gain and index saturation dynamics on a sub-picosecond timescale. The model accounts for bi-directional propagation of ultrashort optical pulses through the amplifier as well as free-carrier absorption and two-photon absorption.
The influence of in-band crosstalk on the error performance of all optical networks with different topologies is studied. A statistical crosstalk model is used for evaluating the bit- error rate. The model accounts for optical preamplification. We present a network topology having the best performance while using the largest transmission path.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.