Pilotless Channel Estimation Scheme using Clustering-based Unsupervised Learning

In a mobile wireless communication system, a pilot signal is essential for channel estimation, but consumes wireless resources to transmit data. If the channel can be estimated using only the data signal without transmitting the pilot signal, efficiency of radio resource utilization for data transmission can be maximized. In this paper, we propose a channel estimation scheme without any pilot signal, which is so called pilotless channel estimation (Pf.E), The PCE only utilize the data signals and K - means clustering algorithm. Although there is no pilot signal overhead in the PCE, mean square error (MSE) and bit error rate (BER) performance of the PCE outperform those of the least square (LS) channel estimator. For practical performance verification, we use the resource allocation unit used in LTE and 5-th generation mobile communication system. In addition, we examine whether PCE can operate in various carrier frequency and various subcarrier spacing according to numerology of 5-th generation mobile communication.