A Survey on Machine and Deep Learning for Optical Communications

The ever-growing complexity of optical communication systems and networks demands sophisticated methodologies to extract meaningful insights from vast amounts of heterogeneous data. Machine learning (ML) and deep learning (DL) have emerged as frontrunners in this domain, offering a transformative approach to data analysis and enabling automated self-configuration in optical communication systems. The adoption of ML and DL in optical communication is driven by the exponential increase in system and link complexity, stemming from the introduction of numerous adjustable and interdependent parameters. This is particularly evident in areas like coherent transceivers, advanced digital signal processing, optical performance monitoring, cross-layer network optimizations, and nonlinearity compensation. While the potential benefits of ML and DL are immense, the extent to which these methods can revolutionize optical communication remains largely unexplored. Additionally, many ML and DL algorithms have yet to be deployed in this field, highlighting the nascent nature of this research area.

ISAR imaging of space objects using encoded apertures

A major threat to satellites is space debris with their low mass and high rotational speed. Accordingly, the short observation time of these objects is a major limitation in space research for appropriate detection and decision. As a result, these objects do not fully illuminated, leading to their incomplete images at any snapshot. In this paper, we propose a method to decrease the number of snapshots in a given observation time and using a limited number of spot beams per snapshot called the encoded aperture. To recover the space debris images, an inverse problem is defined based on compressive sensing methods. Also, we show that for satellite imaging the T V norm is more appropriate. We develop a procedure to recover space debris and satellites using L1 and T V norms. Using simulation results, we compare the results with the well-known SBL and SL0 norm in terms of the number of snapshots, MSE, SNR, and running time. It is shown that our proposed method can successfully recover the space objects images using a fewer number of snapshots.