用於同頻同時全雙工傳輸數位干擾消除演算法的設計與實作

Abstract

通訊技術不斷的演進，從以前的1G到現在的4G，網路速度越來越快，在資訊的取得上越來越便利，5G的概念亦已被提出，其速度會比目前的100Mbps還要多上十倍以上，要達到此目的必然要在傳輸方法上做一些改變，提升頻譜的利用度是達到這個目的的方法之一。 近期許多研究指出同時同頻全雙工傳輸能夠有效率的提升頻譜效益，但是全雙工傳輸會有嚴重的自干擾訊號從傳送端漏給接收端，導致全雙工系統無法有很好的效果，本論文主要在提出如何去消除自干擾訊號讓全雙工傳輸系統能正常運作，並且有接近兩倍的傳輸量。 我們在資料的處理上發現放大器會讓訊號產生諧波，也就是放大器與一些內部原件具有記憶效應，因此我們使用Dynamic Regression建構出記憶效應的模型，讓Digital Interference Cancellation演算法根據傳送的訊號重建出最貼近的接收訊號。 設計出來的演算法會在兩個平台上做驗證，首先會在MATLAB的模擬ad9361的虛擬平台驗證，另一個在LabVIEW的software-defined radio上建構同時同頻全雙工系統，再看看演算法的效果是否能夠讓傳輸的判斷不會出錯。

For enhancing spectrum efficiency, in-band full duplex technology is one of the candidate technical solution. But, self-interference (SI) signal causes dramatically degrade the performance of in-band full-duplex system. In practically, amplifier regards as memoryless system; however, memory effect which cause harmonic in amplifier will be hard to be estimated. Consequently, memory effect of amplifier is considered in this work. Therefore, we utilizes the dynamic regression to implement digital self-interference cancellation (DIC) for mitigate the memory effect. Two platforms are constructed to evaluate this DIC mechanism; one is MATLAB simulation, and the other one is software-defined radio (SDR) platform. In both platforms, the SI signal could effectively reduction and the residual signal could be attenuated close to noise floor. In SDR platform also proves that the proposed DIC algorithm is sufficient to attenuate the SI signal, where successfully perform video streaming demonstration.