Smart Control of OBFN-based Phased Array Systems (SCOPAS)

Background

Photonic processing of phased array antenna (PAA) systems on an optical beam forming network (OBFN) chip has several advantages. Processing in the photonic domain reduces beam squint, can lead to small, robust, and cost effective devices. However, the accurate control of the many on-chip optical ring resonator (ORR) delay elements is difficult as the properties of these elements can vary significantly due to fabrication errors, chip inhomogeneity’s, and heater cross talk. In this project we develop algorithms to control the ORR delay elements based on the measured OBFN signal. Our approach is based on modelling the thermal dynamics and coupling of the ORR in the OBFN within the new framework of decomposable distributed systems, pioneered by prof. Michel Verhaegen. A data-driven OBFN control algorithm is developed which identifies, from acquired data, the thermal coupling and local dynamics of the ORR. This model and its uncertainty is used in devising a robust control for high performance OBFN systems. With the new decomposable framework it becomes possible to design the global control of an OBFN possibly consisting of thousands of ORRs with the complexity of the control design for a single ORR only. Furthermore, the new approach enables a distributed implementation of the controller on a GPU. The models and developed algorithms are implemented, tested, and validated on an OBFN test assembly from the company LioniX.

Project team members

ir. Laurens Bliek
dr. Sander Wahls
prof. Michel Verhaegen

Keywords

Data-driven control, satellite communication, online optimization, surrogate modeling, machine learning

Partners

LioniX International

Goal/ objective

In this project, smart control algorithms for on-chip optical beam forming network (OBFN) based phased array antenna (PAA) systems are developed. These algorithms are essential for the high quality and flexible performance of OBFN-based PAA systems. The developed algorithms are tested on an OBFN test assembly from SATRAX and, as a result, the developed algorithms are readily implementable on the first generation of OBFN-based PAAs. The successful implementation of these algorithms impacts the whole development line of OBFN-based PAAs as the availability of adaptive photonic OBFNs relaxes design constraints and fabrication tolerances and creates more flexible OBFN systems that have higher performance. Consequently, the algorithms are an essential step in the development of high performance OBFN-based PAA systems to be used by end-users such as Astron, Astrium, KLM and NLR. Moreover, it keeps the LioniX OBFN systems at the forefront of the electronic beam-steering antenna systems market.

Work programme

The OBFN system will be investigated and modelled, so all algorithms can be tested on a simulation of the system. The OBFN system consists of various subsystems for which physical models are available. The simulation will be verified by conducting experiments on the real system that is available at LioniX.

A data-driven control algorithm will be designed to control the OBFN in a simulation. Again, these results will be verified on the real OBFN system. The algorithm should be able to work in an online setting where more and more data becomes available, and it should be able to deal with measurement noise and model errors.

Outcome

A data-driven control algorithm called Data-based Online Nonlinear Extremum-seeker (DONE) has been developed during this project. It is able to automatically tune the ORRs in the OBFN system to steer the antenna in the desired direction.

L. Bliek, H. R. G. W. Verstraete, M. Verhaegen and S. Wahls, "Online optimization with costly and noisy measurements using random Fourier expansions", IEEE transactions on neural networks and learning systems, 2016. DOI: 10.1109/TNNLS.2016.2615134

Selected publications