透過干擾管理與影片位元率選擇之改善自適性影音串流於效用感知

Abstract

視訊串流是一種非常消耗網路資源的服務，在多重基地台的環境下，基地台間彼此的干擾（稱作細胞間干擾）會造成網路頻寬下降，為了維持視訊串流的服務品質，基地台間的資源管理及干擾抑制是非常重要的，不過只有提升頻寬是不夠的，還需要針對訊串流的應用層特性做考慮。自適性影音串流是近年來最常被使用的視訊串流技術，主要的概念是透過客戶端自行監測網路狀況來挑選適當的影片片段品質做播放，然而在自適性影音串流的演算法中，影片品質挑選僅是由客戶端自己選擇，而且也只透過應用層的參數做決定，沒有來自實體層的資訊，若是能透過伺服器端輔助，則更能適應於當下的網路狀況。在這篇論文中，我們綜合地考慮這兩個層面並提出一個整合式的干擾管理與影片位元率選擇策略來改善自適性影音串流服務於一個有細胞間干擾的環境，並且實作於長期演進技術（LTE）的實測平台上分析其表現。表現衡量是透過一個效用分數函數來做量化分析，結果顯示我們的方法在各個情境下皆大幅提升了自適性影音串流的效用分數。

Video streaming are now becoming a major service in wireless network. However, video streaming is a very consuming service that occupies lots of network bandwidth. In Long Term Evolution (LTE) multi-cellular network topology, the resource reuse factor is usually set to 1, which lets adjacent base stations use same frequency band to increase resources efficiency. However, User Equipments (UE) at cell edge may suffer from high interference from neighbor cells. The shortage of network bandwidth caused by the interference from the neighbor cells becomes an issue called Inter-Cell Interference (ICI). Inter-Cell Interference Coordination (ICIC) was introduced to solve ICI problem by 3GPP [3]. To maintain service quality, the management of radio resources and interference mitigation are crucial for multi-cellular networks. To improve video streaming services, increasing network bandwidth is not enough to enhance video streaming service. Application layer characteristics will be also considered. Dynamic Adaptive Streaming over HTTP (DASH) is a popular video streaming technique in recent years. The main concept is that DASH service clients monitor the network conditions and select the video segments with proper quality. However, the network conditions considered in client-based DASH algorithm are usually application level parameters, such as buffer filled level, rebuffering time, historical throughput, etc. There is no physical layer information in DASH application. If the selection of the video quality can be aided by the physical channel information from server, it will more adapt to current network situation. In this thesis, we jointly consider this two aspects and propose a integral interference management and rate selection strategy procedure to improve DASH service in ICI environment. The performance of our design was verified by real LTE testbed system. The quantification of performance analysis is by a DASH utility score function [25]. The results showed we largely improve the DASH utility in each scenario.