Research On Computational Imaging Based Environment Sensing System

Sensing has become a key feature and technology in 6G communication networks.Different from previous generations of communication networks,sensing will break traditional design principles,completely change the architecture of traditional communication networks,and change the way of interacting with the communication environment.How to sense and reconstruct the communication environment,so as to further improve the efficiency of the communication system,is a huge challenge that needs to be solved.The main contents of this paper are as follows:First,this paper proposes a mm Wave antenna array based computational imaging system.In order to better design and optimize the communication scheme,the base station usually needs to obtain accurate channel state information of all users in the area,so channel estimation is an important part of the wireless communication system.In general,channel estimation is based on statistical channel models.However,since the wireless channel is determined by the propagation mechanism of electromagnetic waves(such as reflection,scattering,diffraction,etc.)and the real physical scattering environment(such as the scatterers in the scene,the layout of the reflector,etc.),if we can make full use of the inherent information of the environment,actively sensing communication environment to establish a true geometric model,then we will transform the channel from a statistical model to a deterministic model,thereby greatly improving the efficiency of communication.In order to solve this problem,we propose a mm Wave antenna array based computational imaging system,which is an intra-beam super-resolution method based on the idea of random radiation and stochastic modulation of wavefront.This method is formed by modulating the wavefront of the detection signal into a temporal-spatial coherent radiation field,and then obtains the target information in the beam through the calculation between the echo signal and the detection signal,thereby constructing a deterministic model of the environment.The simulation results show that this method can effectively image and reconstruct the environment.Second,this paper proposes distributed computational imaging with reconfigurable intelligent surface.In previous computational imaging systems,the transceiver is an antenna array and each antenna element needs to be modulated separately,which requires many radio-frequency chains and complicated signal processing techniques.In addition,the traditional method uses colocated transmitting and receiving antenna which is not conducive to the flexibility of the system.Reconfigurable intelligent surface(RIS)is a revolutionary technology for achieving spectrum and energy efficient in future wireless communication systems.In this paper,we propose distributed computational imaging with reconfigurable intelligent surface.In this system,the transmitting antenna,receiving antenna and RIS are distributed in different spatial positions.By deploying reconfigurable intelligent surface,the phase and amplitude of the detecting signals are stochastically modulated by reconfigurable passive elements of RIS,which achieves high spectrum and energy efficiency with low hardware cost.Furthermore,we analyze four propagation models of the detecting signal in detail and subdivide region of interest into a collection of grids to formulate the reconstruction process as a sparse recovery problem.Simulation results demonstrate that the proposed distributed computational imaging with RIS can achieve outstanding imaging performance and reconstruct the target effectively.Third,this paper proposes an adaptive resolution environment sensing algorithm.When building the model,we divide the continuous target space into discrete spatial grids,and set the center of the scatterer as the center of the spatial grid.However,since the scatterers in reality are distributed in a continuous space,the center sometimes lies outside the grid,and we cannot know the size and distribution of the target to be detected in advance.When the resolution,that is,the size of the grid,does not match the size of the scatterers in the actual scene,off-grid problem occurs.If we want to solve this problem simply by further dividing the space grid,as the resolution increases,the grid becomes more refined,the coherence between the measurement matrices increases.Since many compressive sensing algorithms rely on the low coherence of the measurement matrix,it will be difficult to simply use traditional compressive sensing algorithms for sparse recovery.Therefore,the algorithm proposed in this paper uses the idea of band exclusion and coherence band to alleviate the off-grid problem and high coherence of the reference matrix.And by calculating the residual to control the iteration depth,we can find the appropriate resolution and get the reconstruction results.Simulation results demonstrate that the proposed environment sensing algorithm can effectively reduce reconstruction errors and improve imaging quality.