Microwave Photonics brings unique advantages in terms of low loss, high bandwidth and immunity to electromagnetic interference. The use of Photonic Integrated Circuits in Microwave Photonics allows a reduction in the footprint and power consumption by integrating as many components as possible in a single chip. In addition, Microwave Photonics enables key processing features, such as fast tunability, which are very complex (or even impossible to achieve) using electronic approaches.
Programmable photonics adds a new advantage to Microwave Photonics systems: full reconfigurability and multitasking capabilities. The different functions that can be emulated in the programmable mesh include the possibility of configuring any kind of filter and delay line, along with multiplexing/de-multiplexing signals. All these functions together in the same chip allow different applications such as frequency discrimination, frequency up&down conversion, phase shifting, RF filtering, optical beamforming for phased-array antennas, antenna routing, optoelectronic oscillation, arbitrary waveform generation and instantaneous frequency measurement. Programmable photonic technologies have a big potential to reduce power consumption and scale down infrastructure in base stations.
Programmable Photonics opens a new era for fiber communications, enabling the development of highly efficient devices that will reduce power consumption and scale down infrastructure. The capabilities of the device include filtering, add-drop wavelength multiplexing, switching and other functionalities such as fiber network sensing. The possibility of performing different processes using a single chip brings novel applications such as a fully programmable transceivers enabling dynamic optical bandwidth allocation, providing an elastic bandwidth to serve the application/service components generating variable traffic volumes.
Programmable devices enable point-to-multipoint transmission, broadcasting and multicasting, working as an interconnect, providing for the first time the Software Define Network control plane with total flexibility to allocate resources. This reconfigurable network solution circumvents the limitations of static networks by reducing cabling costs and the need to over-provision links. In addition, it is a scalable and modular solution that enables grow/pay-as-needed approaches, activated in a license-based fashion. Furthermore, dispersion compensation and equalization of the imbalance in responsivities for coherent detection can be done on-the-fly using the programmable mesh discharging the DSP of the most time-consuming tasks and reducing power consumption by performing the signal processing in the optical domain.
Machine Learning has become almost omnipresent in daily life in applications such as efficient speech and handwriting recognition, artificial intelligence in computer games, computer vision or image classification. We expect computing systems will become central in a broad range of future applications, e.g., self-driving cars or automated medical diagnostics.
Programmable Photonics hardware has a tremendous potential to impact in all these applications. Case in point, the most computationally intensive part in a neural network and blockchain mining are the matrix-vector multiplications (MVMs). Key properties of a MVM are its bidirectionality (ability to perform the multiplication with the transposed matrix in the opposite direction) and trainability (ability to adapt itself to the forward and the error signal). Programmable photonics using iPronics SmartLight architecture enables both features by using feedbackward signal propagation, flexible interconnection schemes and programmable delays. In a similar fashion, it would be possible to adapt the design and materials to use single photon laser sources allowing for the use of programmable photonics as quantum gates. These features enable alternative computing hardware implementations, such as recurrent neural networks, reservoir and quantum computing.
Programmable photonics finds application in the sensing field expanding the capabilities of multisensing platforms. In this case, the device will act as an interconnect device able to interrogate different kinds of transducers, such as biosensors and fiber-based sensors for structural applications, as well as spectrometers for chemical sensing. In the transmitter part, the array of lasers usually employed for multiplexing (for instance, for detection of different molecules in a single sample) could be replaced by a cheap broadband frequency comb laser. In this way, the operation wavelength to interrogate the sample can be selected in the programmable photonics mesh including linewidth reduction as well. In the receiver part, the programmable mesh can perform, for instance, filtering or dispersion compensation, and can act as well as an interrogator by adding arrayed waveguide gratings as building blocks.
Programmable photonic processing also allows beamforming for optical phased arrays, essential functionality in emergent fields such as LIDAR.
