OTFS系統中基於分離領航塊的序列性通道估測

Abstract

為了實現數據決策(data decision)，延遲-都卜勒(delay-Doppler, DD)網格上的通道矩陣(delay-Doppler channel matrix, DD-CM)估測對於在延遲-都卜勒通道上運行的正交時頻空間調變(orthogonal time frequency space modulation, OTFS)系統至關重要。然而，準確的DD-CM估測是具有挑戰性的，因為DD-CM是由接收DD網格上的路徑響應(path responses)、延遲位移(delay shifts)和都卜勒位移(Doppler shifts)非線性組成的通道響應(channel responses)所構成。本文提出了一種序列性通道估計(sequential channel estimation, SCE)方法，通過分離傳輸的均勻間隔的領航符元塊，依序估測延遲位移、路徑響應和都卜勒位移，然後間接組成DD-CM估測。當通道在接收DD網格上表現出整數值的延遲位移和實數值的都卜勒位移時，SCE方法可以對個別通道參數(individual channel parameters, ICP)進行精細估測，從而精確估計DD-CM。由於精細的ICP估測，SCE方法提供了比現有的幾個通道估測(channel estimation)方法更精確的DD-CM估測，這些方法從分離的領航幀(separate pilot frames, SPF)或從混合數據/領航幀(hybrid data/pilot frames, HF)中提取通道訊息。為了應對延遲-都卜勒通道在延遲和都卜勒方面的廣泛色散，SCE方法比所有從HF或SPF中提取DD-CM的通道估測方法需要更少的領航負載(pilot overhead)。

To enable data decision, the estimation of the channel matrix on the delay-Doppler grid (DD-CM) is essential to the orthogonal time frequency space (OTFS) systems operating over delay-Doppler channels. Accurate DD-CM estimation is, however, challenging to attain in that DD-CM consists of channel responses composed nonlinearly from path responses, delay shifts and Doppler shifts on the receive DD grid. Aided by separately transmitted blocks of uniformly spaced pilot symbols, a sequential channel estimation (SCE) scheme is proposed in this report to estimate delay shifts, path responses and Doppler shifts sequentially and then compose the DD-CM estimate indirectly. The SCE scheme can provide fine estimation on individual channel parameters (ICP) and thereby estimate DD-CM accurately when the channel exhibits integer-valued delay shifts and real-valued Doppler shifts on the receive DD grid. Due to fine ICP estimation, the SCE scheme provides more accurate DDCM estimation than several existing channel estimation schemes which extract channel information from separate pilot blocks or from hybrid data/pilot frames. To combat with the delay-Doppler channels with wide dispersion ranges in delay and Doppler, the SCE scheme requires less pilot overhead than all channel estimation schemes which extract DD-CM from hybrid data/pilot frames or from separate pilot frames.