LTE下鏈路非連續接收機制中支援服務品質資源分配演算法

Abstract

長程演進技術(LTE)為第三代合作伙伴(3GPP)發展第四代行動通訊過程中形成的無線網路系統。長程演進技術採用正交分頻多工存取 (OFDMA)做為下行鏈路實體層的傳輸技術，其頻譜多樣性提供很好的網路效能。為了有效地使用無線資源，目前已經發展出許多不同目標的資源分配演算法。由於現今智慧終端普及，對於多媒體應用上網的需求呈現高度成長，保證即時性訊務的服務品質成為封包排程必須要克服的問題。 由於長程演進技術是藉由許多新穎技術構成的網路系統，使用者裝置必須擁有較複雜的計算能力增加了電力消耗。因此第三代合作伙伴針對下行鏈路提出非連續接收模式(DRX)來降低電力耗損，延長使用者裝置的電池壽命。使用者裝置在非連續接收模式中不再永遠保持清醒地監聽下傳控制通道(PDCCH)。當使用者裝置進入睡眠模式時會關閉射頻(RF)電路，停止接收來進演進型基地台(eNB)的訊息，以達到省電的效果。雖然非連續接收模式擁有很好的省電效能，但是其代價便是增加了封包延遲。因此在非連續接收模式中保證即時性訊務的服務品質變得更加困難。傳統的資源分配演算法不需要考慮使用者裝置會進入睡眠模式的問題，因此只在當前的訊框中滿足訊務的需求。而在非連續接收模式中，資源分配必須考慮使用者裝置可能會在睡眠模式違反服務品質需求，況且現今尚未有一種演算法是針對在非連續接收模式中做資源分配。因此我們提出感知非連續接收模式的資源分配演算法降低非連續接收模式對訊務造成的影響。[1]利用訊務之延遲限制與資料遺失率計算『最小所需頻寬』，將資源分配定義為滿足訊務之最小所需頻寬下，最大化系統吞吐量之最佳化問題。而在非連續接收模式中，計算最小所需頻寬將改為預留服務品質策略，用來推算使用者裝置是否能夠進入睡眠模式。針對那些進入睡眠模式後會違反服務品質需求的使用者裝置，我們會動態地增加啟動非活動計時器(Inactivity Timer)的機會，防止它們進入睡眠模式並繼續保持傳輸。模擬結果顯示我們提出之資源分配演算法不但能夠降低非連續接收模式對系統吞吐量和平均遺失率的影響，還能夠保持良好的省電效能。

The UTRAN LTE specifications are developed by 3GPP. LTE adopts OFDMA as its downlink physical transmission technology which provides high user diversity and frequency diversity that improves network performance. In order to effectively use the limited wireless resources, there are many resource allocation algorithms which have been developed for guarantee of different targets. But nowadays, smartphones are very popular so that the multimedia applications are growing rapidly. Therefore, guaranteeing the QoS requirements of real-time traffics is a challenge for resource allocation algorithms. LTE is composed of many novel technologies which cause more needs of UEs’ computationally complex capability that drains its battery quickly. Thus, 3GPP proposed DRX mode for LTE downlink transmission to extend the UE’s battery life. When UEs configure DRX mode, it is no longer for UEs to keep listening to PDCCH all the time. During sleeping period, UEs shutdown RF circuit and stop receiving information from eNB to achieve power saving. Obviously, the trade-off of DRX mode is the increased packet delay. Hence, in DRX mode, to guarantee the QoS requirements of real-time traffics will be more difficult. The conventional resource allocation algorithms try to satisfy the requirements of the UE only during the basic dynamic scheduling time-domain unit and they do not take the effects of DRX mode into account. Therefore, we propose two DRX-aware resource allocation algorithms which decrease the effects in DRX mode. [1] exploits traffic delay bound and loss probability and running loss probability to calculate minimum requested bandwidth, in which resource allocation is defined as a problem of maximizing system throughput while the minimum requested bandwidth of traffic is guaranteed. In DRX mode, the calculation of the minimum requested bandwidth will adopt QoS provisioning policy to evaluate if the UE will violate its QoS requirements after entering sleeping mode. For those UEs which should not be allowed for entering sleeping mode, we exploit INACTIVITY-TIMER that makes them keep awake and prevents the violation of QoS requirements. Simulation results show that our proposed DRX-aware resource allocation not only decreases the harms of system throughput and the packet loss rate which are caused by DRX mode but also achieves good power saving performance.