Mitigating Channel Power Leakage in Millimeter‐Wave MIMO Systems Using an Optimized Physically Consistent Neural Network

ABSTRACT The growing demand for high‐speed wireless communication has driven significant advancements in millimeter‐wave (mmWave) technologies, particularly in next‐generation systems like 5G and beyond. Millimeter‐wave Multiple‐Input Multiple‐Output (MIMO) systems have emerged as a promising solution to address the need for increased data rates, capacity, and spectral efficiency. This manuscript proposes a novel approach named Mitigating Channel Power Leakage in Millimeter‐Wave MIMO Systems Using an Optimized Physically Consistent Neural Network (MCPL‐MW‐MIMO‐PCNN). Here, transmitting as well as receiving beamformers are chosen via offline training of an analog beamforming (ABF) network, which takes the channel as input. To mitigate channel power leakage, the Physically Consistent Neural Network (PCNN) is employed, aligning the channel gains of selected beams in the same direction to maximize the received signal‐to‐noise ratio (SNR). Finally, the Multi Objective Fitness Dependent Optimizer (MOFDO) is considered to optimize the weight parameter of the PCNN classifier, which accurately mitigates channel power leakage. The proposed MCPL‐MW‐MIMO‐PCNN is implemented in Python. The performance of the MCPL‐MW‐MIMO‐PCNN approach attains 18.36%, 22.58%, and 29.99% lower bit error rate and 23.58%, 24.36%, and 30.72% higher spectral efficiency when compared to the existing techniques: DL dependent analog beam forming design for mmWave massive MIMO scheme (ABD‐MWMMS‐TNN), DL driven hybrid beam forming technique for mmWave MIMO system (HBM‐MWMS‐ANN), and DL framework for beam selection and power control in M‐MIMO‐mmWave communications (BSPC‐MIMO‐LSTM) methods.