Project

In this research project, we propose a number of novel major breakthrough contributions to the construction of a super high-speed on-chip integrated information network based on controllable silicon photonic via AI that has the potential for applications of photonic neural networks and processing big data as well as other high-speed processing applications.

The project achieves the overall goals on the intellectual merit because of importantly major activities. Firstly, a proposal for the construction of ultrahigh-speed on-chip photonic communication network with some advantages of low loss, full-mesh configuration, flexible connectivity, which based on silicon photonic wave-guide technologies. Secondly, the project proposes novel structures that play the heart role of photonic networks are optical cross-connect chip and optical add/drop multiplexing chips. This is a worthy intellectual property because of the originality of the silicon photonic based-optical cross-connected devices that have not yet been published before. Finally, the proposal also succeeds in applying AI techniques, especially deep learning, to the management, control and reconfiguration of the photonic networks-on-chip.

The project proposes a prototype via experimental fabrication of a photonic integrated circuit with optical wave-guide signal cross-connect function or optical mode multiplexing/demultiplexing function by mean of modern manufacturing technology (such as DUV photo-lithography 193 nm or e-beam lithography), which results in the sample image being captured with SEM (scanning electron microscope). Characterizing the optical performance of transmission via an infrared wavelength spectrum (window region 1550 nm of optical communication). The broader impacts of this prototype is to make a development platform for many applications such as 5G, data center, big data processing, deep learning as well as make an extremely enormous commercial value, for example, only one of widely optical switch market is worth up to 11 billion US dollars. Moreover, this is the first time a prototype of high-tech products has been created in Vietnam, which contributes to the development of the photogenic VLSI manufacturing industry that has the market value up to many billion US dollars.

The results of planned project in details will contribute to top-ranked scientific research through at least the valuable publications of four high-ranking and peer-reviewed scientific papers and one reputation conference article. It is direct contributions to the improvement and development of the advanced research for the scientific community in Vietnam and the world.

In this research project, with the main activities listed above, we propose some new contributions to the construction of super high-speed integrated information networks based on silicon photonics and the use of deep learning models to control, manage and configure the network. In addition, the project contribution is also the foundation for the application of photonic neural networks and large data processing applications based on photonic network infrastructure on ultra-wide bandwidth chips, small size, and high integration capability, low loss and the ability to integrate monolithic on the same chip. Improving the technology to build these infrastructures can create functional classes that act as "neurons" for ultrafast signal processing in the photonic neural network or for the processing nodes with big data.

[Activity.1]: Research on integrated optical networks on reconfigured chips is using silicon waveguides, the way to create inter-chip, in-chip information networks, how to create large-scale photonic circuits from the small-sized primary photonic circuits. Research on deep learning and application capability in photonic network-on-chip.

[Activity.2]: Propose a new photonic circuit structure to handle optical channel signal to construct a broadband wavelength multimode division multiplexing device with controllable switching capability in the configuration of 1+1 or 1:1 to protect ultrahigh capacity optical communication routes and optical communication device monitoring capability and the ability to expand the number of signal modes.

[Activity.3]: Propose new 2×2 switches for simultaneous broadband optical signal modes. Proposing the new structure of the interleaved split-mode multiplexing device or the configuration wavelength for applications that alternate channel flow, channel selection, and multicast connectivity.

[Activity.4]: Propose a new structure of the basic optical channel crossover structure for 3 and 4 ultra-wide bandwidth ports using mode conversion technology, thereby building on-chip integrated network topology with the ability to expand by block multiplexing or pair with Banyan diagram.

[Activity.5]: Investigation deep models and photonic neural networks simulating the human brain, photonic neurons, ability to connect, learn, and memorize information through photonic neuron cells. Investigate the feasibility of the application of deep models in controlling on-chip information resources, connectivity control, optimal management of connection traffic, and the ability to smartly and fast reconfigure with the increase of information network gate sizes.

[Activity.6]: Applying deep learning models to manage and reconfigure the network connections in the network with high efficiency and fast in chip-integrated photonic networks using silicon photonics technology.

[Activity.7]: Write the final report of the whole project.