Modeling And Signal Detection Of MIMO Spatial Energy Detection

The development of next-generation wireless communication networks is driven by high reliability,high-speed mobile communication and the growing demand for user experience.MIMO technology is one of the important technologies in future wireless communication systems,which greatly improves the information transmission rate and spectrum resource utilization of wireless communication systems.Space Multiplexing MIMO based on energy detection has the advantages of insensitivity to Doppler frequency shift and low receiver complexity.However,square-law processing brings nonlinear mutual interference problems to MIMO space multiplexing signal detection.Based on the mutual interference mechanism of multiple signals in nonlinear MIMO,this thesis conducts theoretical research on nonlinear numerical solution,nonlinear dimension expansion,model linearization and nonlinear detection based on deep neural networks.Equivalent linear models based on complementary multiband amplitude waveform and energy difference detection,and low-complexity linear detection algorithms are derived and designed.A deep neural network detection method for nonlinear mutual interference models is also developed.The former requires estimation of real channel information,while the latter does not require pre-estimation of channel information,but requires supervised training of the network.Problems such as nonlinear multiplicative interference,noise resistance and reduced interference ability were solved,efficient MIMO spatial modulation and detection methods with low hardware complexity were obtained,and the theoretical system and practical methods of noncoherent MIMO spatial separation were improved.The main contents of the thesis are as follows:(1)Firstly,the research background and significance of noncoherent MIMO energy detection system and deep neural network detection are described,and the research status of noncoherent MIMO technology and deep neural network detection technology at home and abroad.Then,three types of noncoherent MIMO system models based on energy detection are introduced,namely,the nonlinear system model of noncoherent MIMO MASK energy detection,the quasi-linear model with the addition of actual array elements,and the linear model with the addition of virtual array elements.This thesis analyzes the modelling and detection theory of MIMO energy detection,providing a theoretical basis and reference for the linear modelling and detection and nonlinear detection methods presented in this thesis.(2)In response to the nonlinear mutual interference problem between spatial channels caused by noncoherent MIMO energy detection,a multi-amplitude complementary waveform is designed,and the system is equivalent to a linear space division system model using energy difference detection.This solves the problem of the dimensional expansion of the channel matrix caused by the energy detection of the received signals in noncoherent MIMO,as well as the problem of the solution of the error rank matrix.Based on MIMO-MASK,MIMO-FSK and chirp-BOK modulation and detection,this thesis designs a MIMO dual-frequency multi-band complementary modulation scheme that can achieve multi-band modulation capability with low detection complexity.A ZF-ML joint detection algorithm is proposed based on a linear space division system model.(3)This thesis presents a signal detection scheme based on deep neural networks for nonlinear cross interference models.The system model and specific algorithm implementation process are given,and the conditions for neural networks to solve linearized models are analyzed.Finally,through numerical simulation analysis,it is shown that the deep neural network detection method based on the nonlinear cross interference model has a good bit error rate performance,which can effectively improve the reliability of MIMO-MASK communication systems.Furthermore,the bit error rate performance of different system models with Doppler frequency shifts under different detection algorithms is compared and analyzed,which proves that the proposed detection methods have good anti-Doppler characteristics.