An Iterative PRISMA Review of Deep Learning, Sparse Bayesian, and Compressed Sensing Methods for Channel Estimation in Modern Wireless Systems

Accurate channel estimation is basic to efficient functioning of modern wireless communication systems, particularly in the case of 5G, mmWave, MIMO, and vehicular networks. However, most conventional techniques are usually affected by the complexity of such methods, pilot overhead, and limitations in high mobility or dynamic environments. On the other hand, in spite of a large number of proposed methods, there is a gap in covering a comprehensive and systematic survey to compare several techniques under various performance metrics. Most of the existing review articles focus on very limited sets of models and do not critically compare or contrast models based on their efficiency, accuracy, and computational complexity. To fill these gaps, this paper bases its review on state-of-the-art channel estimation models PRISMA, and included methods span traditional signal processing approaches such as LMMSE, adaptive frequency-domain estimators, and compressed sensing-based techniques, and prominent approaches based on deep learning frameworks including CNN and hybrid estimators. With the PRISMA methodology, this review provides a comprehensive analysis of the methods involved in various applications, such as MIMO, mmWave, RIS-aided systems, and underwater communications. The proposed review, on giving a number of models, discusses the pros and cons of those models. Models based on deep learning bring in an improved accuracy along with the potential of improved computational performance in estimation, whereas sparse and hybrid models have shown impressive performances in heavily resource-constrained environments such as underwater acoustic and high-mobility vehicular communications. This work will prove to be very informative for the researcher and engineer, both in showing optimal models for various applications of communication and in laying down the basis for future integration of GAN and advanced machine-learning methods into channel estimation. The outcomes derived from this review will shape and inform the future design and deployment of wireless systems through the development of a strong platform for harnessing optimization strategies in diverse applications of channel estimation.