A Wireless Propagation Model Based on Artificial Intelligence Technology

with the rapid development of artificial intelligence (AI), machine learning technology also has the ability to perform online operations. Since electromagnetic waves are greatly affected by environmental factors during the propagation process, it is easy to generate complex scattering and diffraction during the propagation process. The existing wireless propagation models need to be further corrected by using a large amount of collected engineering data in practical applications, so a wireless propagation model based on machine learning need to be built, and the problem of the rational deployment of base stations can be solved. In order to obtain the propagation model that accords with the actual environment of the target region, this paper collects a large amount of additional measured data, engineering parameters and electronic maps to correct the propagation model after learning which is based on the traditional wireless propagation model function fitting process under the condition that the engineering parameters, geographical location information, and RSRP (average signal receiving power) of a specific geographical location measurement point are known. After appropriate features are be designed, the feature set will be performed by using divergent analysis, neural network connection fitting and principal component analysis. Then the effective eigenvalues are selected and input into the BiLSTM-based wireless propagation model. BiLSTM is used to characterize the relationship between features and the relationship between the characteristics and the target value RSRP. The BiLSTM representation relationship will be output through a layer of fully connected output. After continuous training and tuning, the RSRP of different geographical locations is predicted.