A Low Complexity Channel Estimation Technique for NB-IoT Downlink System

3GPP introduced Narrow-Band Internet of Things (NB-IoT) in release-13 with a special feature to work with only 180 kHz bandwidth. Effective channel estimation is highly important for adequate receiver performance of NB-IoT system. Linear Minimum Mean Square Error (LMMSE) technique is very effective for estimating the channel condition but possesses high complexity. Singular value decomposition (SVD) and splitting the channel autocorrelation matrix into several submatrices reduces the complexity of LMMSE technique. In this paper, we propose a modified low complexity and computationally efficient LMMSE estimator by linking the advantages of both techniques stated above with overlap banded technique in channel autocorrelation matrix for NB-IoT downlink (in-band) system. In the proposed technique, subdivided channel autocorrelation matrices are overlapped among them and hence reduces complexity. Simulation results show that by dint of negligible degradation of performance, the complexity is significantly reduced.