基於決策原則之叢發正交分頻多工於車用通訊的低延遲通道估計

Abstract

物聯網(IoT)時代的來臨，帶動汽車通訊的技術發展與需求，車用通訊議題勢將在未來更加熱門。而正交分頻多工(OFDM)技術是現今通訊技術主流且持續應用於未來的第五代行動通訊標準(5G)上，包括車用通訊等。而IEEE 802.11p是以OFDM為主的一個的現存重要車用通訊標準。而車用通道估計因其通道具有高時變性的關係是車用通訊裡其具挑戰的議題。現存的通道估計方法缺乏足夠的精確度或是需要較大的延遲(latency)或額外引導(pilot)以取得足夠的精確度。我們提出一種低延遲且具有高精確度的方法來解決此問題，並對此方法作深入之探討。本論文考慮的系統是具有前置資料(preamble)的OFDM架構。這個設計的主要部分是藉由相鄰OFDM符元(symbol)具有時域關聯性(temporal correlation)以決策定向(decision-directed)的方式來求取通道。並試著在相對簡化的通道上，分析其效能以了解其特性並驗證正確性。其後再利用OFDM符元中的引導做頻域(frequency domain)上的遞迴式濾波以求取更佳的精確度。再者，藉由多重接收天線下的助益，在效能上做近一步的加強，展現出空間分集(space diversity)的優點。最後模擬驗證上，單一接收天線與多重接收天線情形分別透過被用於驗證效能的車用通道上在IEEE 802.11p上來展現此方法的優越性。此研究提供在未來車用通道估計上可行改進之方向。

We consider bursty orthogonal frequency-division multiplexing (OFDM) signal transmission where each signal burst (or packet) consists of a preamble followed by data symbols which may contain few pilots. Channel estimation in vehicular environments for such signals is a challenging issue, as is evidenced by related studies for the IEEE 802.11p standard. Existing channel estimation methods often yield deficient performance or require relatively high latency to achieve better performance, unless additional pilots are introduced. We propose a technique that addresses this problem. The primary aspect of the design is to exploit the temporal correlation in channel responses over close-by OFDM symbols in a decision-oriented manner. A secondary aspect is to take advantage of any available pilots in the data symbols in a frequency-domain filtering built on the temporal processing results. Regarding the performance of the proposed method, we first analyze the proposed temporal processing technique under some simplistic channel conditions both for insight and for verification. Further simulation results based on the IEEE 802.11p specifications under a measurement-based fading multipath vehicle-to-vehicle channel model show the superiority of the proposed method.