-
StatusCompleted
-
Status date2019-07-09
-
Activity Code5C.012
The main objective of the project was to develop and demonstrate fast and high-throughput (Tb/s) optical switching technology that could be applied for optical packet routing in a burst-switched processor in a telecommunication satellite. The envisioned switch architecture was consolidated and the required key components were developed. Finally, a breadboard demonstrator was designed and fabricated. There were some technical faults in the final breadboard, but if those would be fixed the full implementation of the envisioned optical 110x128 switch should clearly fulfil all the target specifications and it would be even scalable into a much larger switch (>>1 Tb/s).
The objective of this activity was to design, manufacture and test a representative end-to-end breadboard of an ultra-fast & high-throughput optical switch. This optical switch would ultimately be the core part of an ultra-fast & high-throughput burst-switched processor for meshed-type satellite repeaters. The development of the complete processor was not the subject of this activity. The optical switch of the complete processor should have at least 100 input channels and 100 output channels that should be connected to each other in arbitrary order and with a switching time of ~1 ns. The breadboard assembled in OTUS should have less channels, but it should have the same switching time and it should be scalable to the full implementation of the optical switch (100x100 minimum).
This activity was also expected to include the delta technology development of the critical optical components of the optical switch. The most critical component to be developed was an optically tuneable wavelength filter (or selector) on a silicon-on-insulator (SOI) waveguide platform with a response time of ~1 ns. Also optical transmitters, wavelength multiplexers, optical fiber amplifiers, passive splitters and hybrid integrated detectors were to be either procured or developed in the project.
The originally envisioned architecture was not possible to implement, so changes to the project plan were necessary from the beginning. The development of the key components also suffered from many technical problems that were tackled. This led to significant delays in the project. Finally, the breadboard could not be tested as a whole because of some faults in the wavelength selector chips, especially in the AWG operation and in the high-speed operation of the PDs.
The results of the project should enable the realisation of an ultra-fast & high-throughput burst-switched processor that uses transparent burst switching, channel multiplexing, and bandwidth asymmetry. The proposed architecture will provide a higher satellite throughput, with respect to a full regenerative processor approach, as well as simple signal broadcasting. Only demodulation and decoding of the burst header (i.e. label) is necessary, instead of the demodulating and decoding the complete burst (i.e. burst header and burst payload).
The technology development to be carried out in the project is a prerequisite for the successful realisation of the proposed burst-switched processor with an optical core. However, the results of the project are expected to find applications also in other space projects, as well as outside the space sector. The obtained results show that with optoelectronics integration on Si it is possible to significantly reduce the footprint, weight and power consumption of optical systems, such as those studied in this project.
The breadboard and the envisioned full switch were based on the "broadcast-and-select" architecture illustrated in the figure below. This architecture provides strictly non-blocking switching between any input and output channel, and the additional benefit of simple signal broadcasting. A unique wavelength channel is reserved for each of the M input ports. The number of output ports should be somewhat larger than the number of input ports (e.g. 100x128) to reduce latency problems in practical operation. The required key components are:
- Lasers (10 Gb/s, covering the entire C-band (~1550 nm) with 50 GHz channel spacing, procured),
- Wavelength multiplexer (procured),
- Fibre amplifier (flat spectrum, low cross-talk between wavelengths, procured),
- Passive signal splitter (procured),
- Fast wavelength filters/selectors (response time ~1 ns, developed on SOI with hybrid integrated optoelectonics),
- Detectors (10 Gb/s, broad spectral range, developed and integrated on SOI).
The breadboard contained two wavelength selector chips and all the relevant components along the optical path from one laser to those chips. The wavelength selector chips (WSCs) had monolithically integrated arrayed waveguide gratings (AWGs) and star couplers based on SOI waveguides, as well as thermo compression bonded semiconductor optical amplifiers (SOAs) and photodiodes (PDs).
Schematic overview of the complete optical switch based on the "broadcast-and-select" architecture.
The project consortium has the potential to provide all the key components needed to construct the complete optical switch, instead of procuring most of the required parts. The plan was to increase the integration density by developing and optimising most of the key components, which will reduce the weight and power consumption of the optical switch. The backup plan was to procure commercial components when needed. The target was to design a full switch and to implement a breadboard that demonstrates the ability to reach all the target specifications for the full switch.
The project has been completed.
A technology roadmap has been proposed towards the full implementation of the switch and for enabling even further improvements and scalability of the switch reaching far beyond the 1 Tb/s target for the aggregate data traffic through the switch.
