Research On Beamforming And Nonlinear Interference Cancellation Algorithms In Low-orbit Broadband Satellites

With the increasing demand for long-distance,seamless coverage,high-speed,and low latency communication,traditional high-orbit satellites can no longer meet the communication needs.Therefore,low orbit multi beam satellites have become a research hotspot in the next generation of wireless communication technology.In low orbit multi beam systems,there are two core issues.One is how to suppress signals in interference directions while generating point beams with strong directionality;One is how to effectively process the data streams within each beam to minimize inter stream interference caused by beam sidelobes.Problem 1 relies on beamforming algorithms.Most existing beamforming algorithms require iterative gradient descent algorithms to search for antenna weights,resulting in high computational complexity and large computational delay,which affects the real-time performance of communication;However,algorithms with lower computational complexity have unstable suppression of interference directions and poor performance.Problem 2 relies on the precoding algorithm,which can process the data flow of each beam based on the interference channel state information between beams and eliminate interference between users.However,existing precoding algorithms suffer from low energy efficiency and high computational complexity,and a precoding scheme that balances performance and complexity needs to be studied.In response to problem one,this article proposes an autoencoder based beamforming algorithm(ABB),which can learn the phase characteristics of the received signal itself.Compared with traditional RBF network beamforming,it can significantly reduce the fitting dimension,thereby reducing the network size.At the same time,it can reduce the occurrence of overfitting and has strong universality.Simulation shows that ABB has interference suppression capability close to optimal beamforming,and its suppression effect on interference signals is 20dB enhanced compared to traditional beamforming algorithms such as minimum mean square error.At the same time,as the direction of the signal and interference changes,there is no need to retrain the network,greatly reducing the algorithm delay.In response to question two,this article proposes a List encoding of Vector Perturbation Precoding(LVP)method based on a multi encoder architecture.The LVP algorithm adopts a parallel iterative search architecture with multiple encoders.Each encoder selects one of the many possible paths in each iteration based on appropriate path selection criteria,and retains it until the next iteration.After all iterations are completed,the historical optimal encoding vector will be used as the output result of the algorithm.The simulation results show that the LVP algorithm can reduce the probability of falling into local optima while preserving the low complexity of the low dimensional search algorithm.On the premise of achieving the same BER performance,compared with traditional ball search based(Schnorr Euchner VP,SE VP),the computational complexity is reduced by more than 2 orders of magnitude;Compared with similar low dimensional search algorithms such as the Dimension-2 Vector Perturbation(D2VP),the complexity is reduced by half.In addition,the LVP algorithm provides a new solution for the Closest Point Problem in grid search problems.