Macrocell及Femtocell共存系統的頻道分配最佳化方法設計

Abstract

有鑑於femtocell與macrocell共存系統是於將來行動通訊的趨勢之一，本論文研究主題是femtocell及macrocell共存系統的頻譜分配最佳化機制設計。論文情境設定為一個擁有頻譜執照的無線服務供應商，並且擁有macrocell及femtocell系統，所考慮的是該如何分配有限的頻譜資源供femtocell及macrocell接取以達到最佳的使用者服務品質。 研究首先探討目前此種共存系統的頻譜分配方式之優劣，我們採用“分割式頻譜再用(split spectrum reuse)”之分配方式做為本論文的最佳化頻譜分配方法。分割式頻譜再用法將無線服務供應商所擁有的頻帶切割為兩部分，一部分專供macrocell接取，另一部分則專供femtocell接取。其次深入討論以分割式頻譜再用方式在提供語音服務方面，femtocell及macrocell的最佳頻道數分配計算模型與方法。 我們採用話務阻隔機率做為頻道分配最佳化的效能指標，阻隔機率表示用戶使用無線傳輸時無法建立連線的機率。我們將各個基站建模單一為M/M/c/c系統，為了分析femtocell及macrocell基站的阻隔機率，我們需先計算各基站的話務流量（到達率）以及話務流量佔用頻道的時間（服務率之倒數）。然而無線服務用戶具有移動性，用戶會在各個基站之間進行話務的轉接，造成分析上的困難，因此我們針對此種特性分別對macrocell及femtocell基站之話務流量及頻道佔用時間建模以求解到達率及服務率，並以嘗錯法（try and error）來最佳化頻譜分配以達最低的話務阻隔機率。 針對以上所提出之頻道分配最佳化方法，以現存之無線通訊網路系統為基礎，進而設計動態頻道分配最佳化模組RANSAM （Radio Access Network Spectrum Allocation Module），藉以在現有femtocell及macrocell共存系統上以切割式頻譜再用方式進行動態頻道分配最佳化，其可根據系統中的使用者行為的變化來動態決定femtocell及macrocell的最佳化頻道分配數目，使系統的使用者服務品質最佳。 我們使用MATLAB軟體來進行模型的模擬與最佳化。藉由MATLAB軟體的模擬結果，我們驗證了femtocell幫助macrocell分流的果效，缺少femtocell來協助分流將使得macrocell的話務阻隔機率最高。我們也發現了不同頻道分配數目對系統的影響，不適當的頻道分配將造成高話務阻隔機率，當分配過多頻道給femtocell時，將使得macrocell的頻道不敷使用；而當分配過少頻道給femtocell時，將造成femtocell協助macrocell分流的成效不彰。這些實驗結果驗證了femtocell與macrocell之間存在一最佳的頻道分配數目，因此我們針對此頻道分配數目最佳化，並發現影響最佳化頻道數目的關鍵因素流量；當系統整體流量越高，macrocell越依賴femtocell幫忙分流以降低macrocell的阻隔機率，因此需要分配越多頻道給femtocell基站使用。

Femtocell and macrocell coexistence network is a trend of mobile communications. Our research focuses on optimal spectrum allocation for the coexistence network. The scenario we consider is a mobile service operator, who owns spectrum license and operates a femtocell and macrocell coexistence network. The operator would like to allocate limited spectrum channels for accessing femtocell and macrocell to reach maximum network performance. Two common channel allocation mechanisms are compared: “split spectrum reuse” and “shared spectrum reuse.” We choose the former one as our allocation mechanism in which channels are separately used by femtocell users and macrocell users. This thesis focues on how to decide the optimal number of channels to allocate among femtocells and macrocell. We adopt call blocking probability as the performance indicator, which indicates the probability that a call request is denied. Each cell is modeled as a M/M/c/c system. To analyze blocking probabilities of femtocell and macrocell respectively, we need to compute each cell’s call arrival rate and call service rate. User mobility and handoffs among cells lead to a challenge of analysis. To address this challenge, we establish models of channel occupation and call transfer among cells to facilitate the analysis. Furthermore, “try and error” is adopted to optimize the numbers of channel allocation for maximal system performance. To implement the optimization scheme, we design RANSAM (Radio Access Network Spectrum Allocation Module) to allocate channels between macrocell and femtocells dynamically depending users’ traffic. Through MATLAB software simulations, the effect of sharing macrocell’s loading by femtocells has been verified. Without femtocell, the macrocell has a high call blocking probability. We have also verified the effect of different split ratios for channel allocation; inappropriate split ratios would cause high call blocking probabilities. In the one hand, too many channels allocated to femtocells will cause the shortage of macrocell channel; on the other hand too many channel to macrocell will cause the lack of femtocell’s traffic sharing effect. Simulation results demonstrate the existence of an optimal channel split ratio. Further stydy shows that when system traffic grows, the optimal number of femtocell channel also increases because the macrocell needs to rely on traffic sharers (femtocells) more.