藉由多重訊號分類演算法及壓縮感知做盲通道估測

Abstract

毫米波為第五代行動通訊中一重要技術，它使我們有更高的資料傳輸率以及更大的可使用頻寬。然而，由於毫米波的超高頻率特性使其受限於極高的途徑損失以及比起傳統無線通訊更少的路徑數目。熱門的解決辦法為採用大規模多輸入多輸出及波束形成技術。而我們需要通道狀態資訊才能利用數種已存的電波束形成器及結合器設計演算法。在絕大多數的通道估測研究中，研究者們傾向採用領航序列來增加估測的準確性，但這會降低資料傳輸率。因此我們試圖提出一個可以避免使用領航序列以及維持可接受的估測準確率的新穎方法。我們進一步地將這個問題分為兩種情況。第一種情況是單一路徑，也就是每一個使用者到基地台的通道都為只有一個。多訊號分類演算法以及巢狀陣列將被用於此情況。而另一種情況則為多路徑。我們利用壓縮感知以及巢狀陣列來估測通道。

Millimeter wave (mmWave) cellular communication system is a promising technology in the fifth-generation communication. Because of the ultra-high frequency, it suffers from forbiddingly high path loss and the paths from users to the base station (BS) are much fewer than the traditional wireless communications systems. It is popular to exploit massive multi-input multi-output (MIMO) systems and beamforming techniques to deal with this problem. The pilot-based channel estimation algorithms become more inefficient when the number of antennas increases. Therefore, we propose a novel method that avoids using the pilot sequence. We first estimate the user number by the rank of the correlation matrix of the received signals. Then, the compressed sensing technique is utilized to estimate the direction-of-arrival (DOA) of each path. Eventually, we take advantage of the special structure of the fading matrix to eliminate the spurious DOAs and estimate the fading gain of each path at the same time. A special case that the available path from each user to the base station (BS) is one is also discussed, and we exploit multiple signal classification (MUSIC) algorithm to reduce complexity. In both cases, a non-uniform antenna array called the nested array is utilized to increase the array aperture.