Search

In this letter, we mainly focus on the significant multipath component identification issue for orthogonal frequency division multiplexing (OFDM) wireless communication systems in a fast time-varying scenario with a low signal-to-noise ratio (SNR). Considering the time-domain channel prediction demand of OFDM systems, we introduce the joint singular spectrum analysis (JSSA) to identify the significant multipath components of the channel impulse response (CIR) by investigating the local predictability of the multipath component in detail. The JSSA has two parts, i.e., the random SSA (R-SSA) and the predictable signal reconstruction (PSR). The former is used for multipath component decomposition, while the latter is used to reconstruct the multipath component based on the local predictabilities of those subcomponents. Therefore, the JSSA can reduce the negative effect of noise and solve the significant multipath component identification issue. The simulations indicate that the JSSA in our letter has good identification performance in the given communication scenario.

Frequency division duplex (FDD) downlink massive multiple-input and multiple-output (MIMO) systems encounter difficulty in acquiring the sparsity and threshold parameter. As a remedy, a time-frequency block sparse channel estimation method based on compressed sensing (CS) is proposed that consists of the generalized block adaptive matching pursuit (gBAMP) algorithm. By using the time-frequency joint block sparsity of massive MIMO systems, the index sets selected in the iterative process of the algorithm are further optimized to improve the stability of the algorithm. Then, in the absence of the threshold parameter, the adaptive iterative stop condition is determined using the residual based on the Frobenius norm (F-norm), and the effectiveness of the method is proved. Simulations show that the algorithm can quickly and accurately estimate the massive MIMO channel information and outperforms its counterparts.