A Novel Millimeter Wave Channel Estimation Algorithm Based on IC-ELM

The millimeter wave (mmWave) communication with high frequency bands can improve the capacity of the wireless network significantly. However, the large bandwidth and time varying characteristic of the mmWave channel lead to a large increase in the computational complexity of conventional channel estimation algorithms. In this paper, we proposed a novel mmWave channel estimation algorithm based on Imperialist Competitive-Extreme Learning Machine (IC-ELM). The number of hidden neurons is optimized by Imperialist Competitive Algorithm (ICA) in the structure of Extreme Learning Machine (ELM) according to the mean square error (MSE) between the actual and estimated channel state information (CSI). The IC-ELM is trained by the channel frequency response (CFR) of pilot positions to learn the channel characteristics. Further, the CSI of mmWave channel can be estimated by the trained IC-ELM network. Compared with conventional channel estimation algorithms, the simulation results show that the proposed IC-ELM mmWave channel estimation algorithm can achieve better performance in terms of MSE and bit error rate (BER). And the proposed IC-ELM is available in different types of mmWave channel models.