OFDM無線系統下有限封包長度之機會式群播排程研究

Abstract

蜂巢式網路(Cellular Networks)中的基地台(Base Station)需要滿足不同使用者的服務需求，在每一個時程假設每一位使用者對於頻道的SNR狀況呈現I.I.D.，基地台必須選擇適當的頻道(Channel)傳輸速率。採用機會式排程(Opportunistic Scheduling)方式並依據當時情況，讓基地台規劃每一次頻道的傳輸速率，以符合大多數使用者需求，增加系統效益。 在正交分頻多工(Orthogonal Frequency Division Multiplexing)系統中，我們套用Integer Programming概念在I.I.D.環境下逼近最佳解，基地台可以依照使用者該次時程的通道狀態資訊(channel state information)，了解使用者目前可接受的最高傳輸速率為何。基於有限封包長度(finite length)，假設使用者收到足夠的編碼封包，就能夠還原訊息，完成使用者的需求。我們設計出一套演算法可讓基地台依據使用者封包的收取狀況制定優先權，優先權高的使用者群即可決定此次時程的頻道傳輸速率。 最後經由實驗模擬結果顯示，此演算法可讓整體系統在每一個時程能夠彈性地服務和選擇適當傳輸速率，另外不管是在多少人數環境底下，整體系統花費的時程以及整體系統throughput都呈現很好的效益。

We consider opportunistic scheduling for multicasting stream fragments over OFDM downlink channels in a single cell. Assume that channel states are i.i.d. and available at the base station through uplink channel feedback. Furthermore, assume that each finite length message fragment has been erasure-encoded into packets such that the user is able to decode and obtain the original fragment when receiving no less than K packets. Under such a system and assumptions, it is important for the opportunistic multicast scheduler to determine an appropriate instantaneous channel rate to achieve the shortest time required to ensure all users able to decode a sending fragment. We resort to a weighting function to divide the set of multicast users into two subsets. One subset contains users whose services are critical for system performance. The other subset contains the rest. Each channel transmission rate is then determined from the critical subset of users. We study the effect of the partition factor for dividing the set of users on the system performance through simulations.