Maximum‐likelihood synchronization and channel estimation with multiuser detection in GFDMA

Abstract

Accurate estimation and correction of channel distortions and carrier frequency offset (CFO) are of a great importance in any multicarrier communication system. Hence, in this paper, we propose data‐aided CFO and channel estimation techniques for both multiuser uplink and downlink of the generalized frequency division multiple access (GFDMA). Our proposed solutions jointly estimate the CFO and channel responses based on the maximum‐likelihood criterion. To simplify the implementation of the proposed estimation algorithms, we suggest a preamble composed of two similar Zadoff‐Chu training sequences in a generalized frequency division multiplexing block. It is worth mentioning that our proposed technique can estimate both integer and fractional CFO values without any limitation on the acquisition range of CFO. In the uplink phase, each user aligns its carrier frequency with the base station using the estimated CFO in the downlink. However, the CFO estimates may get outdated for the uplink transmission. Thus, residual CFOs may still remain in the received signal at the base station. While being trivial in the downlink, CFO correction is a challenging task in the uplink. Thus, we also propose a joint CFO correction and channel equalization technique for the uplink of GFDMA systems. Finally, we evaluate our proposed estimation and correction algorithms in terms of estimation mean square error and bit error rate performance through simulations.

Low‐complexity search method for CFO estimation in GFDM

“…long enough to accommodate the wireless channel delay spread and use a root-raised cosine prototype filter with the roll-off factor of a = 0.1, in all the simulations [6]. 10 10 N CM method in [5] method in [6] method in [7], (γ = 1) proposed method, γ = L/2 proposed method, γ = L Fig. 2 Computational complexity comparison for different methods Fig.…”

“…As shown in this figure, the MSE performance of our proposed estimator improves as signal-to-noise ratio (SNR) increases and is close to the relevant Cramer-Rao bound. We also compare our method with the ones in [5,6] where the PN and ZC training sequences are deployed. Although our proposed method leads to the same MSE performance as the solution in [6], our proposed method has a lower computational complexity than the one in [6].…”

“…However, all the estimation methods in [1][2][3][4] are limited to the CFO acquisition range of only half subcarrier spacing. To address this issue, a few solutions in the literature have been proposed [5][6][7]. Li et al [5] proposed a joint CFO and channel estimation method which can be straightforwardly extended to GFDM.…”

“…Consequently, the search space will be reduced to a small number of search regions. Therefore, we propose a search method that leads to around an order of magnitude complexity reduction without any performance penalty compared with the solutions in [5][6][7]. We demonstrate superiority of our proposed method over the solutions in [5][6][7] through computational complexity analysis and performance comparisons using simulations.…”

“…System model: Let M and N be the total number of GFDM symbols and subcarriers in each GFDM block, respectively. According to the frequency spreading GFDM transmitter structure in [6], the GFDM transmit signal can be constructed as…”