Abstract—A comprehensive model and simulation framework for an adaptive wireless communication system aided by Intelligent Reflecting Surfaces (IRS) are presented in this research. A feedback-driven learning technique based on neural network back propagation is used to dynamically optimize the IRS system. The behavior of the system is described by a number of equations, such as gradient-based IRS updates, power modelling, Signal-to-Interference-Plus-Noise Ratio (SINR), and Resulting Bit Error Rate (BER). The performance gains in SINR and BER under various power, Reflection Coefficient (R_C), and IRS panel count circumstances are shown. The presented learning model iteratively updates IRS reflection settings until convergence by adjusting in real-time based on measured system performance. This work integrates theoretical modelling, simulation data. IRS-assisted wireless communication system presented in this paper is driven by a Stochastic Gradient Descent (SGD) adaptive learning algorithm, without requiring explicit channel state information, the suggested model dynamically adjusts the IRS phase shifts to maximize the SINR in real time, while the BER is evaluated as a performance indicator. The obtained results show that the presented approach performs better than current methods in terms of spectrum efficiency, SINR gain, and convergence rate.  
 
Keywords—Intelligent Reflecting Surfaces (IRS), stochastic gradient descent, Signal-to-Interference-Plus-Noise Ratio (SINR), Bit Error Rate (BER), wireless communication, adaptive learning