在彈性雲端無線接取網路中佈建無線寬頻頭端設備考量通道容量有限的前端回程網路

Abstract

在第四代行動通訊演進漸趨成熟之下,學界業界紛紛開始尋求下一代行動通訊的新技術以 支撐未來爆量的需求。雲端無線接取網路 (C-RAN) 被視為一個相當有前景的解決方案,由集 中基頻運算池 (BBU pool) 、無線寬頻頭端設備 (RRH) 和前端回程網路 (Fronthaul) 所組成。然 而,雲端無線接取網路對於前端回程網路的頻寬需求極大,若是前端回程網路無法負荷集中基 頻運算池和無線寬頻頭端設備之間的基頻資料,則會對雲端無線接取網路的表現有嚴重的影響, 而目前所知的資料壓縮技術並無法有效率地解決這個問題。 根據集中基頻運算池和無線寬頻頭端設備之間的不同運算功能的切分與擺放,常見的模式 有兩種:完全集中以及部分集中,完全集中是把所有的運算功能都放到集中基頻運算池內,而 部分集中則是把第一層基頻運算功能留在無線寬頻頭端設備上。兩者相比,所需的前端回程網 路頻寬可以相差到 20 到 50 倍之多。 所以,這篇論文的目標是提出一種彈性的雲端無線接取網路架構,此架構中含有兩種不同 集中程度的無線寬頻頭端設備,類似前面所提到的完全集中以及部分集中。我們的問題是要在 眾多的候選地點中選出一群子集來分別佈建這兩種無線寬頻頭端設備,並考量通道容量有限的 前端回程網路的限制以及根據平均的使用者的需求決定要怎麼連接及怎麼佈建。原問題的複雜 度相當高,所以我們接著依照此問題的特性提出一個低複雜度的貪心演算法,透過演算法證明 這個架構的表現確實優於其他兩種基準架構(完全集中和部分集中),以及不同的無線寬頻頭 端設備有各自適合的使用情境。

As the deployment and commercial operation of 4G systems are speeding up, technologists worldwide have begun searching for next generation wireless solutions. Cloud radio access networks (C-RAN), which is composed of three main components: BBU pool, fronthaul and RRH, has been thought of as a promising solution. However, the massive fronthaul bandwidth required to aggregate baseband samples from RRH to BBU pool has a significant impact on the performance of C-RAN and existing baseband compression algorithms can hardly solve this issue. According to different function splitting and placing between BBU pool and RRH, there are two kinds of C-RAN solutions: one is called “full centralization”, where layer 1 functions and beyond are located in BBU pool; the other is called “partial centralization” or “flexible C-RAN”, where the RRH integrates not only the radio function but also some of the baseband functions (e.g. L1 functions), while all other higher layer functions are still located in the BBU pool. Compared with the “fully centralization”, the RRH-BBU pool connection of “partially centralization” only need to carry demodulated data, which is only 1/20~1/50 of the original baseband I/Q sample data. So, our target of this work is to propose that a new flexible C-RAN architecture that there are two types of RRHs with different degree of centralization. Both of them have different pros and cons. We want to select a subset of candidate sites to install these two types of RRHs and to assign demand nodes to the available one taking into account the traffic demand, deployment costs and limited fronthaul capacity. Because the original problem is extremely complicated, we proposed a greedy algorithm with low complexity and small computation time to solve the problem sub-optimally. Then, we run simulation to show that our proposed algorithm performs well then other two benchmark schemes and through the simulation, we figure out some idea about different type of RRH are suitable for some special cases. To the best of our knowledge, we are the first article proposed this architecture and discussing about how the deployment should be in the flexible C-RAN.