20 May 2020 in IEEE Transactions on Industrial Electronics
This paper proposes a new digital signal recovery (DSR) technique for next-generation power efficient low Earth orbit (LEO) satellite-to-ground communication systems, which feature additive white Gaussian noise (AWGN) channel and significant power variation. This technique utilizes the prior knowledge (i.e., nonlinearities of radio frequency power amplifiers (RF-PAs)) of space-borne transmitters to improve the quality of the signal received at ground stations by modeling and mitigating the imperfection using deep neural networks (DNNs). Benefiting from its robustness against noise and power variation, the proposed DNN based DSR technique (DNN-DSR), can correct high signal distortions caused by the nonlinearities and hence allows RF-PAs to work close to their saturation region, leading to a high power efficiency of the LEO satellites. This work has been validated by both simulations and experiments, in comparison with the power back-off technique as well as memory polynomial based DSR solutions. Experimental results show that the DNN-DSR technique can increase the drain efficiency of the space-borne RF-PA from 32.6% to 45% while maintaining the same error vector magnitude as the power back-off technique. It has also been demonstrated that the proposed DNN-DSR technique can handle a signal power variation of 12 dB, which is challenging for conventional solutions.
14 December 2020 in IEEE Access
This paper proposes a novel deep neural network (DNN)-based receiver for next-generation low Earth orbit (LEO) satellite communications. The DNN receiver can concurrently compensate for multiple imperfections of the satellite communication system to improve the quality of satellite-to-ground transmission. A special focus has been placed on handling the nonlinear distortion in the transmitted signal caused by space-borne high-efficiency radio frequency power amplifiers (RF-PAs), which is crucial in high-throughput satellite communications, but has been overlooked by existing relevant research. In this receiver, a DNN is designed and trained to learn the channel effects, nonlinearities of the RF-PAs, and digital modulation schemes in the received signal for demodulation and nonlinearity/channel effect compensation at the same time. The proposed receiver has been evaluated using five popular filtered orthogonal frequency division modulations with the nonlinear distortions experimentally extracted from a real gallium nitride (GaN) RF-PA and the additive white Gaussian noise channel generated by simulations. The validation results demonstrate that the DNN receiver can accommodate different modulation schemes and two typical groups of RF-PA classes with a satisfactory bit error rate performance. It has the potential to boost the performance of existing on-orbit LEO satellite communication systems with minimal system modifications and serves as a promising solution for future satellite communication services.
5 October 2021 in IEEE Access
This paper proposes a robust digital signal recovery (DSR) technique to tackle the high signal-to-noise ratio (SNR) variation and transmitter memory effects for broadband power efficient down-link in next-generation low Earth orbit (LEO) satellite constellations. The robustness against low SNR is achieved by concurrently integrating magnitude normalization and noise feature filtering using a filtering block built with one batch normalization (BN) layer and two bidirectional long short-term memory (BiLSTM) layers. Moreover, unlike existing deep neural network-based DSR techniques (DNN-DSR), which failed to effectively take into account the memory effects of radio-frequency power amplifiers (RF-PAs) in the model design, the proposed BiLSTM-DSR technique can extracts the sequential characteristics of the adjacent in-phase (I) and quadrature (Q) samples, and hence can obtain superior memory effects compensation compared with the DNN-DSR technique. Experimental validation results of the proposed BiLSTM-DSR with a 100 MHz bandwidth OFDM signal demonstrate an excellent performance of 11.83 dB and 9.4% improvement for adjacent channel power ratio (ACPR) and error vector magnitude (EVM), respectively. BiLSTM-DSR also outperforms the existing DNN-DSR technique in terms of the ACPR and EVM by 2.4 dB and 0.9%, which provides a promising solution for developing deep learning-assisted receivers for high-throughput LEO satellite networks.
18 March 2024 in IEEE Transactions on Wireless Communications
Inter-satellite links are significant to achieve complex space missions like high precision ranging and rendezvous navigation. However, the impact of spacecraft multipath on system performance cannot be ignored, especially for large spacecraft. To evaluate and optimize the satellite communication system, a geometry-based non-stationary inter-large-satellite wireless channel model is proposed. It includes three components, i.e., a line-of-sight component, solar-panel-bounce parts for multipath on solar panels, and body-bounce (BB) components resulting from spacecraft body scattering. The BB clusters’ time-variant azimuth angle of arrival and departure are derived because conventional angle distribution functions, like von Mises, are unsuitable for scatterers directionally located in a limited region in satellite scenarios. Channel evolution in the space-time-frequency (STF) domain is modeled with birth-death processes regarding self-rotation and solar panel rotation in addition to three-dimensional (3D) movement. Then, channel characteristics, e.g., STF correlation functions, Doppler power spectrum density, and stationary intervals, are derived and studied. Finally, the proposed channel model is validated to be accurate by ray-tracing simulation, satellite measurement and on-orbit observation.
12 August 2024 in IEEE Microwave and Wireless Technology Letters
This article presents a novel artificial neural network (ANN)-based digital predistortion (DPD) coefficients prediction (ANN-DPDCP) technique for dynamic nonlinearities induced by varying input power levels of power amplifiers (PAs). Conventional DPD techniques face challenges in mitigating dynamic nonlinearities efficiently. By modeling and predicting variations of conventional Volterra-based DPD coefficients using ANNs, the ANN-DPDCP technique rapidly provides appropriate DPD coefficients based on the target input power level. Benefiting from its concise training dataset and fitting capability, the ANN-DPDCP technique requires limited storage resources and derives DPD coefficients at arbitrary input power levels with negligible delay and comparable linearization performance. Experiments on a $Ka$ -band PA driven by 100-and 400-MHz signals with a 12-dBm input power range illustrate storage resource reductions of 99.54% for 400 MHz and 99.81% for 100 MHz.