The optical switching technology gap: Optical switching techniques have been holding promise to comprise this alternative approach, providing a means to flatten the intra-DC networks and scale their size, capacity and overall performance. These expectations have been based on features inherent to optical switches such as data rate and protocol transparency, ultra-low latency, low power consumption and possibility for scaling the switching capacity using different colours and WDM techniques. They have been also based on the possibility under conditions for scaling up the number of ports in optical switches, and the possibility under a different set of conditions for achieving ultra-fast switching times. Despite these expectations, the optical technology has been very short so far in turning this potential into true commercial impact. A possible explanation is the lack of a platform that can bring together these features and provide an optical switching solution with overall superior performance compared to the electrical counterparts.
Figure 1: Reference architecture of intra-DC networks described by Facebook. The network is based on electrical packet switches and its hierarchy consists of 3 layers with ToR (first layer), aggregate (second layer) and spine switches (third layer).
The hybrid integration of PolyBoard and TriPleX platform is aimed to be the key for the development of a novel optical switching technology that will enable devices with large port number, modular size expansion, fast switching time, low polarization sensitivity, low power consumption and low cost. The port number in integrated devices depends on the integration yield and density, and on the possibility to cascade large numbers of elementary switching structures. 3PEAT invests on TriPleX that features high integration density and ultra-low propagation loss (0.1 dB/cm), allowing by default for long cascades. This possibility is used today for the development of large beamforming networks in microwave photonics technology, and can be used in a similar way for large switching fabrics. The waveguide crossings, which represent a well-known problem in integration technology and might be the limiting factor for the fabrication of large-scale PICs, will be eliminated using the 3D PolyBoard platform and the possibility it provides for routing optical lines via different layers. The yield, which is extremely high for simple TriPleX chips, is expected to remain high in the case of chips with PZT elements. Collaboration between the partners LioniX and SolMateS has already resulted in stable deposition process with high yield. Finally, to further extend the number of ports in a modular way without risk regarding the PIC size and yield, 3PEAT will use the optical FlexLines on PolyBoard technology for the interconnection of smaller PICs. In parallel, 3PEAT technology will be also able to provide fast operation with switching times down to 20 ns, meeting the switching speed requirements of Ethernet and InfiniBand networks. The fast switching feature of the technology will be based on the stress-optic effect and the use of the PZT elements inside the MZI switches in the TriPleX part.
Artistic layout of the hybrid 3D PIC for the active switching module (here shown as 4×4). The final PIC will have a TriPleX section with a 36×18 switching matrix based on PZT-based MZIs, and a 3D PolyBoard section with 7 waveguiding layers and 18 AWGs in total. Two such PICs will be interconnected via an Optical FlexLine in order to form the basis of a 36×36 active switching module.
The challenge of compact optical systems for remote sensing and ranging applications: The case of Laser Doppler Vibrometers
Current tradeoff between performance and compactness: Optical systems for remote sensing and ranging applications are attracting enormous interest in a large variety of fields. Either as Light Detection and Ranging (LIDAR) systems using time-of-flight techniques or as Laser Doppler Vibrometers (LDVs) using the Doppler effect, these systems represent a class of optical solutions, which are very complex and demanding in terms of range of supported functionalities. These functionalities involve at first place the generation and modulation of light, the collimation and scanning of optical beams, the collection of reflections from the surrounding environment, the polarization handling of the reflected light, and the detection of the same light using direct or coherent detection techniques. In the case of high-end applications, the relevant systems can become extremely demanding also in terms of noise performance, setting ultimate challenges for the quality of operation of the individual components and units. In order to meet this challenge in terms of multi-functionality and performance, the relevant systems are based on the use of high-end, bulk components and free-space optics-based assembly techniques. In this way, they become extremely expensive and bulky, creating issues in practical cases, where the system has presumably to fit in small physical space or to be compact and tolerant to mechanical vibrations. A powerful photonic integration technology supporting a wide range of optical functionalities and enabling high photonic performance could be the key to resolve these issues. However, this technology is still missing today.
The case of Laser Doppler Vibrometers (LDVs): The LDVs is a representative example of high demanding remote sensing systems. LDVs are used today in various industries and technologies (e.g. automotive, aerospace, constructions, semiconductors and biomedicine) in order to provide surface monitoring and vibration profile extraction without attaching any element onto the surface that could affect the result. High-end commercial systems are able to detect surface displacements in the sub-picometer regime both at low and high vibration frequencies. This performance can be achieved even in the case of surfaces that are strongly scattering or absorbing, or may be placed several 100s of meters away, reducing the intensity of the reflected signal back to the device.
Optical phased arrays for integrated beam scanning units and current limitations: The optical phased arrays (OPAs) have the potential to replace the lenses and the moving mirrors in the beam scanning unit of remote sensing systems, and enable realizations of this unit in an integrated form. Being the optical analogy of the microwave phased array antennas, the OPAs comprise a group of optical antennas (i.e. light emitting elements) with reconfigurable phase and amplitude relations between them [39]. The total optical far-field of this group is the result of the coherent addition of the individual ones, and can be configured a single optical beam. Depending on the placement of the optical antennas along a linear array (1D OPA) or inside a 2D matrix (2D OPA), the OPA has the possibility to adjust the direction of the final beam in one or two dimensions, respectively, by tuning the phase and amplitude relations between the antennas. Important design parameters include the number of antennas in each dimension and the distance between them (pitch).
3PEAT will leverage the hybrid integration platform and will develop a disruptive PIC as the basis of a fully integrated LDV. The PIC will consolidate the innovations of the project in order to overcome the limitations of the current technology and meet the LDV requirements in terms of functionality and performance. It will be based on the 3D PolyBoard technology in order to develop a disruptive 2D OPA with edge emitters at different polymer layers. It will be also based on the consolidation of the PZT technology in order to develop long arrays of phase-shifters and VOAs for the control of the OPA. Finally, it will rely on the development of the external cavity laser on the TriPleX platform as a laser source with ultra-low phase noise for high-resolution heterodyne measurements.
Artistic layout of the 3D hybrid PIC for the main Laser Doppler Vibrometer module. It will include a low-linewidth laser, an optical circulator, polarization handling elements, a PZT-based phase modulator and a 16×16 optical phased-array (shown here as 4×4 for clearness). The optical phase array will be tuned by 256 PZT phase shifters and 256 PZT-based VOAs.