Abstract
The digital equalization of fibre impairments in coherent optical communications allows for ultimate limits of spectral efficiency to be achieved, while reducing the cost and complexity compared to optical based solutions. Digital backpropagation (BP) was proposed recently as a reduced complexity approach for jointly compensating linear and non-linear impairments in single polarization coherent optical systems. Furthermore, polarization division multiplexed quadrature phase-shift keying (PDM-QPSK) has emerged as a promising format to increase spectral efficiency. Here we assess the performance of the simplified backpropagation equalization in both single channel and multiple channel transmission in combination with polarization multiplexing for different dispersion map configurations. It is shown that polarization multiplexing induces a degradation in performance due to cross-phase modulation polarization scattering, the backpropagation algorithm still overperforming linear equalization.

Abstract
The Daphne project has been addressing the adoption of an optical fiber infrastructure for future aircrafts. Beyond the obvious motivation of reduced weight and electromagnetic interference, the availability of a huge amount of bandwidth makes the optical fiber well suited to transport Radio Frequency (RF) signals transparently, while avoiding cumbersome dedicated RF cabling. An integrated optical network may be exploited to transport radio signals from diverse aircraft antennas ranging from satellite/earth communications, collision avoidance, GPS signals for positioning and attitude determination, weather/detection RADAR to corrosion sensors. Such network can also support passenger infotainment and mobile communication services, such as cellular GSM/UMTS/LTE, broadband Wi-Fi (IEEE 802.11) and Ultra-Wide-Band Wimedia/WiGig. Specifically, the optical fiber infrastructure may provide connectivity from external antennas (through remote nodes) to RF transceivers installed in the cockpit and avionics bay (head-end nodes); in the context of the present paper, the transmission of differential GPS signals used to provide aircraft attitude information will be discussed. The use of GPS for aircraft attitude determination has been under discussion for more than 20 years [1]. It consists in performing carrier-phase differential processing of measurements from GPS antennas affixed to the frame of the aircraft, which yields centimeter- or millimeter-level accuracies, provided that integer phase ambiguities are resolved [2]. The attitude algorithm consists in a highly accurate real time kinematic (RTK) technique, given the short baseline distance between antennas, in which the main antenna acts as a Base station and two auxiliary antennas as Rovers. In the present experiment, we used a setup consisting of two-antennas (Base and Rover), which is enough to evaluate the RTK performance. A particular aspect of concern stems from the fact that the transmission of optical signals through a complex optical fiber network is subject to the occurrence of reflections in the multitude of connectors spanning the path between a remote node and a head-end node. Therefore, we will focus our analysis on the performance impact of optical reflections affecting the power level stability of the optical source. © 2012 IEEE.

Abstract
The measured characteristics of a R-EAM, acting as a base station, were used to assess the performance of a radio-over-fiber uplink for possible provision of wireless services within aircrafts. The specific case of deployment of UWB signals was analyzed. We conclude that laser RIN is a performance degradation factor that imposes a limit on the achievable SNR, especially for a zero biased modulator. A laser RIN of -160 dB/Hz would be required in order to avoid the RIN limitation. Additionally, in this case the performance becomes limited by shot noise. Although the zero bias case seems limitative, the R-EAM can be optimum biased using a small battery, which can last for several months, allowing the base-station to operate as a passive device. © 2011 IEEE.

Abstract
In an optical transceiver, the power consumption related to the operation of the laser device takes a significant parcel of the total consumed power. Thus, in optical networks where a large number of transceiver devices are interconnected, e.g. large distributed sensor networks, it is of great importance to reduce this power consumption. In this work an analysis and simulation results are presented regarding the operation of a bias-free vertical-cavity surface-emitting laser (VCSEL) device, which is based on a previously developed model. The impact on bit-error rate (BER) of the increased turn-on jitter due to the bit-pattern and spontaneous emission is considered. A method for mitigating the eye-diagram distortion penalty based on the received signal equalization is also illustrated.

Abstract
In this paper we experimentally evaluate the performance of a fiber supported radio application, namely WiFi IEEE802.11n, using a low power consumption base-station based on a reflective electro absorption modulator and commercial transceivers. We conclude that the proposed setup is suitable to achieve a transmission of 20 Mbit/s over a 4m wireless channel. A theoretical signal to noise ratio analysis is also carried out, based on experimental results. Finally, we evaluate the possibility of employing the proposed setup in a passive optical network, concluding that this solution may not be cost effective. © 2012 IIMC Ltd.

Abstract
Deep learning methods have been shown to be competitive solutions for modulation classification tasks, but suffer from being computationally expensive, limiting their use on embedded devices. We propose a new deep neural network architecture which employs known structures, depth-wise separable convolution and residual connections, as well as a compression methodology, which combined lead to a tiny and fast algorithm for modulation classification. Our compressed model won the first place in ITU's AI/ML in 5G Challenge 2021, achieving 61.73? compression over the challenge baseline and being over 2.6? better than the second best submission. The source code of this work is publicly available at github.com/ITU-AI- ML-in-5G-Challenge/ITU-ML5G-PS-007-BacalhauNet.