Optimizing the Use of Compressive Sensing on Intelligence Reflecting Surface (IRS) Using Convolutional Neural Network (CNN)

In this research, the author proposes the use of Intelligent Reflecting Surface (IRS) to increase the signal achievable rate in blind spot conditions, such as in an elevator. IRS is a smart reflective surface that can change the direction of electromagnetic signal reflection. In this case, the author uses a Convolutional Neural Network (CNN) to optimize the direction of IRS reflections. CNNs can identify patterns and features in data, which in this case is data about signals and the environment around the IRS. By using CNN, the authors were able to develop a model that allows IRS to automatically adjust the direction of signal reflections, which can help in optimizing achievable rate. The aim of this research is to determine the correct direction of IRS reflection and to determine the achievable rate results after using CS and CNN. This research was carried out using Matlab-based simulation using data sets to train and test the system. It is hoped that the implementation of CS and CNN on IRS can increase the signal achievable rate.