以連續干擾消除實現雲接取網路之上行訊號壓縮

Abstract

這份論文著重於雲接取網路的上行。以在雲處理中心使用每使用者連續干 擾消除技術,使得整個系統節省了計算資源以達到系統架構的精簡。為了 充份利用每個遠程無線電頭所收到的無線射頻訊號間的相關性以及上行通 道的容量,我們採用了實際的 Wyner-Ziv 編碼器來壓縮上行訊號。我們提 出了一套流程來接續地壓縮和重建每個使用者的訊息。我們重新整理了在 遠程無線電頭接收到的訊息形式,使其形式二次高斯旁訊息在解碼器問題 來用餘式 Wyner-Ziv 編碼跑器解決。此外,每使用者連續干擾消除技術和 協同壓縮和解碼大大的減低了系統的算複雜度。餘式 Wyner-Ziv 編碼器是 由兩段連續量化器組成,分別為低密度奇偶校驗碼和低密度生成矩陣碼, 可適用於任意分佈的旁訊息。最後,我們畫出了由消息理論得到的速率區 域和實際模擬得到可傳送的速率點對,比較我們的系統能多靠近理論界限。

This work focus on the uplink of Cloud Radio Access Network (C-RAN) [3]. By adopting per-user successive interference cancellation (SIC) at the central processor, the whole system saved much computing power in turns of simplicity of system architecture. To exploit the correlation of the signal received at each Remote Radio Head (RRH) and to fully utilized the fronthaul capacity, we adopt a practical Wyner-Ziv coder at the fronthaul compression. We propose a flow of algorithm to consecutively compress and reconstruct all the users’ messages. We rearrange the mathematical form of the signal at RRH so that it fits the Quadratic Gaussian noisy side information problem, which can be solved by applying the residual-based Wyner-Ziv coder devel- oped in [8]. Besides, the per-user SIC scheme reduces the system complexity significantly compared to joint compression and decoding. The Wyner-Ziv coder consists of two successive quantizer, made of LDPC and LDGM codes and can be applied to arbitrarily distributed side information. Finally, we plot the achievable rate region obtained from information theory and also transmissible region from simulation, and show how much we approach to the theory.