Dominant CIR Tap Identification for OFDM Channels: Adaptive Bootstrapping Approach

Abstract

The next generation 5G and beyond network, which necessitates advanced physical layer design utilizing distributed data and computational resources intelligently with improved context awareness, is expected to support multi-service traffics fundamentally different from the traditional ones. For this network, orthogonal frequency division multiplexing (OFDM) is believed to be one of the promising candidate waveforms where its performance depends on the accuracy of estimated CSI coefficients obtained via the discrete Fourier transform (DFT) method. This method first estimates the channel impulse response (CIR) followed by taking the DFT of the CIR coefficients. In practice, however, such an estimator suffers from performance degradation when the number of dominant CIR taps (i.e., taps with non-negligible amplitudes) is very small compared to the total size of CIR taps. This paper addresses this limitation by first examining the dominant CIR tap identification problem, and then using only dominant CIR taps in DFT based CSI estimation. In this regard, we propose a novel approach to formulate this problem as a signal to noise ratio (SNR) maximization convex problem where its global optimal solution can be obtained with a simple integer based bisection search. The formulated problem depends on the SNR of each CIR tap which is estimated from the received samples of the previous OFDM data blocks using a new and computationally manageable adaptive bootstrapping technique. We carry out extensive simulations to validate the analytical expressions and examine the effects of different parameters including channel stationarity duration and number of reference sub-carriers which are not used during data transmission (i.e., null sub-carriers). Numerical simulations corroborate the relevance of identifying dominant CIR taps for CSI estimation. In a typical long-term evolution (LTE) channel environment, the proposed approach can achieve up to 60% improvement in spectrum efficiency.