在無線寬頻網路上以期望值模型為基礎進行分層多播串流之排程技術研究

Abstract

在無線寬頻網路上，如 WiMAX、LTE，欲進行一對多的串流服務，由於傳輸品質的不固定，要滿足 Qos 及 QoE 的要求就成了很大的挑戰，如何透過資源排程以達到 QoS 或 QoE 的最佳化也成為近來熱門的研究議題。在很多的相關研究中，為了簡化問題，會假設一段資料可以全部成功解碼的機率只有 0 或 1，並限制同一分層的資料必須使用單一的 MCS 傳送。但我們認為若是可以拿掉這些假設及限制，整體的效益可以獲得提升。此外，我們利用跟影片品質有正向關係的有效傳輸量來定義串流服務效益。因此，我們依據資料傳輸成功的機率模型結合有效傳輸量提出了一個用來估計效益的期望值模型，而且不受限於使用單一 MCS。我們也根據此期望值模型設計出一個用於分層視訊多播串流服務的 Coarse-to-Fine 資源排程演算法，這個演算法適用於多數使用 OFDMA 技術的無線網路環境。藉由 Coarse-to-Fine 的設計，可以減少排程所花費的時間。最後，我們透過實驗驗證拿掉上述的假設及限制後，整體的效益確實有獲得提升，以及我們所提出的排程演算法，適用於多種的環境，如 MS 會移動或有多個 BS 的情形。

In a one-to-many streaming service over wireless broadband networks, such as WiMAX and LTE, it is rather challenging to fulfill the needs for Quality of Service (QoS) and Quality of Experience (QoE) requirements due to frequent topology updates and dynamic channel fluctuation. Resource scheduling as a Qos or QoE optimization problem for wireless broadband networks has attracted growing attention. In many related works, the authors assume that the probability of decoding received data correctly is either zero or one, and restrict that transmitting whole data of a video layer with the same MCS for simplifying the scheduling problem. But we consider that the overall utility can be improved if these assumption and restriction are dropped. Furthermore, we use the effective throughput to define the utility of streaming service, due to that there is a positive relationship between effective throughput and video quality. Therefore, a expectation model based on the probability model of data transmission correct rate is proposed for estimating the utility without homogeneous MCS restriction. A coarse to fine grained resource scheduling algorithm based on expectation model is also proposed for multicasting multiple layered videos over OFDMA-based wireless networks. Since the coarse-to-fine design, the cost time of scheduling algorithm can be reduced. Finally, the simulation results show that the performance is improved and our proposed scheduling algorithm is applicable in many scenarios, such as movable MSs and multiple BSs.