Optimization Of Wireless Propagation Model In Indoor Scenes

Along with the development of Internet of things technology and the gradual development of 5G commercialization,users' demand for high quality wireless communication is increasing.The continuous evolution of business types and data diversification,the rapid growth of indoor mobile data traffic,the increasingly complex electromagnetic environment,the application,control and management of various short-distance wireless devices,the need for efficient and stable information transmission and connection.In order to ensure the quality of indoor wireless communication services,it is necessary to obtain a relatively accurate wireless propagation prediction model to deeply understand and accurately evaluate the propagation characteristics and coverage performance of indoor electromagnetic waves.The wireless propagation model is classified according to the principle and scope of application,and the characteristics and representative models of each type of model are explained.the methods and characteristics of indoor wireless propagation models such as ITU-R P.1238 model,logarithmic distance path loss model,Lee model,ray tracking model are introduced in detail.the advantages and disadvantages of empirical model and theoretical model in engineering applications are compared.Indoor wireless propagation loss is closely related to building structure and material,furniture and so on,and the movement of people and objects will also lead to sharp changes in signal intensity in short distance and short time.the corridor,wall,human body and other scenes were selected for CW(continuous wave)testing to study the attenuation characteristics of electromagnetic waves in different indoor environments and to optimize the Lee model.This paper introduces the test site,test plan,personnel allocation,equipment selection and calibration of each scene,arranges the points for attention in the test and the methods of data processing and analysis,and summarizes the indoor wireless test and model optimization.The basic idea.The results of the model optimization are as follows: the ray model is used to predict the path loss of electromagnetic wave in the corridor horizon scene,and the test results show that the three-ray model can describe the true attenuation of electromagnetic wave more accurately;the influence of different human position and number of human body on the received signal intensity is studied by single person and multi-person test,respectively,and the human body loss formula is established;the prediction method of path loss in the non-sight scene with diffraction propagation is summarized in combination with the change of building structure and signal attenuation;the attenuation of a single wall when electromagnetic wave passes through the inner wall of multiple buildings continuously is studied in the multi-wall scene,and the wall isestablished.The relationship between the loss and the number and frequency of the wall modifies the wall parameters of the model.A new prediction model of indoor wireless propagation is constructed by machine learning.Firstly,the plane map of the test site should be accurately drawn,and the information such as the length and obstacles of each propagation path should be obtained after marking the transmitting and receiving points in the map and gridding test.A variety of transmit power and frequency are selected to be tested separately according to the marked position.11 d parameters such as electromagnetic wave parameters,direct distance,diffraction distance,number of walls and incident angle are selected as input variables,and the signal intensity corresponding to the receiving point is used as output variables to construct indoor data sets for wireless propagation prediction.A variety of machine learning algorithms are used to train the random forest best prediction accuracy,suitable for building models.finally,the practicability and accuracy of Lee model,optimization model and machine learning-based model in indoor propagation prediction are compared.machine learning is trained by extracting electromagnetic wave parameters and indoor building information without dividing into different scenarios.It can predict the signal intensity of the whole wireless coverage site at multiple frequencies and power,and reduce the influence of indoor environmental differences and multipath propagation on the prediction results.