Volume 12, Number 3

An Intelligent Channel Estimation Method for Future Mobile Generation


Khaled H. Almotairi, Umm Al-Qura University, Saudi Arabia


The challenges of the future generations of mobile telephony operators are based on the use of previous generation’s connectivity. The signal reconstruction is a technique that requires in most of the cases, the insertion of Pilot Symbols (PS) to ensure identification data reception. The PS are sent through the network and are known by the receiver and the transmitter. Therefore, they occupy bandwidth but contain no information. For Channel Estimation (C.E.), the LTE-A standard adopts inserting a pilot symbol in seven data symbols for each subcarrier. This is reflected in a throughput loss of the transmitted information. This method considers that channel does not vary during the period of slot transmission, but the rapid time variation is one of channel characteristics. We propose in this work an intelligent solution for C.E. in order to reduce bit rate loss while ensuring a good quality. A combination of two techniques of Artificial Intelligence has been proposed: Deep Learning and Fuzzy Logic. The main motivation of such a combination is to incorporate the complementarity of these two techniques to exploit the advantage of each and overcome its limitation. The proposed method is characterized by a non-random initialization of his weights from a fuzzy rule base. The implementation of this smart system into a C.E. system has reduced the loss of throughput and follows the variations of the propagation channel. The proposed Hybrid Deep Learning model provides better results than classical estimators. It has, also, a low complexity and a fast treatment.


Channel estimation, interpolation, LTE-A, Deep Learning, Fuzzy Logic.