Research On High Spectrally Efficient Sparse Code Multiple Access For Satellite-terrestrial Link In Low Earth Orbit Satellite Network

With the rapid development of Internet of Things(Io T),traditional terrestrial networks cannot meet the global wide-area access requirements of Io T in the future.Low earth orbit(LEO)satellite network has the advantages of global coverage,low propagation delay and low cost.It is an inevitable trend to adopt the LEO satellite network as a supplement to the terrestrial network to build LEO satellite Io T and realize the global interconnection of everything.However,due to the influence of multipath fading and Doppler frequency shift of the satellite-terrestrial link of the LEO satellite network,as well as the limitation of the spectrum resource of the satellite-terrestrial network,higher requirements are put forward on multiple access methods,decoding methods,channel modeling of the satellite-terrestrial link in LEO satellite network.In this thesis,we consider LEO satellite Io T as the research background.Aiming at the characteristics of low utilization efficiency of spectrum resources,severe link communication interference and stochastic channel changes of the satellite-terrestrial link of LEO satellite network,high spectrally efficient sparse code multiple access method for the satellite-terrestrial uplink in LEO satellite network is proposed.The proposed method can realize the unit spectrum resource of the system to accommodate more user information to improve the spectrum efficiency performance.The proposed method can also eliminate the inter-carrier and inter-symbol interference to improve the communication reliability of the system.Furthermore,the proposed method can achieve effective stochastic channel modeling of LEO satellite-terrestrial link based on generative adversarial network.The main research contents are as follows:First of all,this thesis is oriented to LEO satellite Io T application scenarios,layered system architecture and integrated network architecture of LEO satellite Io T are proposed.In view of the characteristics of multipath fading and Doppler frequency shift of the satellite-terrestrial link channel of LEO satellite network,the channel model and probability distribution function are studied and analyzed,and perform channel modeling for satellite-terrestrial link of LEO satellite network.Aiming at the satellite-terrestrial uplink of LEO satellite network,the sparse code multiple access method is used to model the system,and the end-to-end communication system based on neural network is studied.Secondly,a sparse code multiple access method based on spectrally efficient frequency division multiplexing is proposed.Aiming at the efficient utilization of the spectrum efficiency of the satellite-to-earth link of the low-orbit satellite network and the demand for multi-user access,a sparse code multiple access method based on high-efficiency frequency division multiplexing under satellite-terrestrial uplink of LEO satellite network is proposed to construct dual non-orthogonal transmission and access system.Under the condition that the information carried by each sub-carrier remains unchanged,the spectrum efficiency performance of the overall system is improved by compressing the sub-carrier spacing.At the receiving end,in view of the influence of the fading channel of the satellite-terrestrial link in LEO satellite network and the inter-subcarrier interference caused by non-orthogonal waveforms,we proposed to use frequency domain processing method to make iterative detection algorithm can be applied in complex satellite-terrestrial link,and combine with logarithmic domain message passing algorithm to demodulate and decode multi-user information.Thirdly,an iterative multi-user detection and decoding method based on interference cancellation is proposed.In view of the influence of LEO satellite channel characteristics such as multi-path fading and Doppler frequency shift of the satellite-terrestrial link of the LEO satellite network,in order to meet the requirements of reliable multi-user communication under a single beam,an iterative multi-user detection and decoding method based on interference cancellation under satellite-terrestrial uplink of LEO satellite network is proposed on the basis of sparse code multiple access based on spectrally efficient frequency division multiplexing.Under the condition that the spectrum efficiency of the system remains unchanged,the soft information iterative method is adopted at the receiving end,through one external iteration and three internal iteration processes to realize the update of function nodes,variable nodes and decoding nodes,while sub-carrier demodulation,multi-user information decoding and parallel channel decoding are carried out to minimize inter-symbol interference,inter-carrier interference and inter-user interference,which enhance the communication reliability of system.Finally,an end-to-end sparse code multiple access method based on generative adversarial network channel modeling is proposed.Due to the random movement state of the user nodes under the satellite-terrestrial link of LEO satellite network,the channel state information will change randomly.In this situation,channel modeling based on mathematical models cannot accurately evaluate th e performance of mobile communication systems.In view of the above problems,this thesis proposes an end-to-end sparse code multiple access method based on generative adversarial network channel modeling under the satellite-terrestrial uplink of LEO satellite network.We adopt autoencoder self-learning multi-dimensional codebooks and multi-user decoder of sparse code multiple access.For the first time,we propose conditional wasserstein generative adversarial network channel modeling with gradient penalty.Through online and offline training,the feasibility of hyperparameters of convolutional neural network sparse code multiple access system is further verified.