彈性調整動態分時長期演進技術之干擾消除

Abstract

在現行的LTE 網路中，為了因應大量的Data 需求和上下傳的樣 貌變化不同，在TDD 的網路中必須針對不同的UL、DL 的traffic 調 整上下傳策略，而衍生出了Dynamic TDD。Dynamic TDD 可以針對 不同形式的UL 和DL traffic 選擇一個符合自己的需求的方式，以求 small cell 中的throughput 達到最大。然而以Dynamic TDD 去應付高密 度佈署的小型基地台效果十分有限，因為小型基地台之間鄰近的地 理位置，使得干擾非常嚴重。本文針對TDD 的網路進行研究；LTE 中的TDD 即為分時進行DL 和UL，在3GPP 的標準中有定義7 種 configuration，其UL 和DL 的順序皆不大相同。主要干擾影響即為一 者DL 另一者UL 的情況，DL 的Power 大小通常大於由手機端發射出 的UL，造成Signal 不夠高而Interference 相對較大以致SINR 下降，在 此情況下SINR 會比兩者同時選擇UL 時還低，在原本其一者DL 另 一者UL 情況下時的UL throughput 將會降低。基地台主要採取的策 略大致有兩類，其一是Traffic Adaptation，但此方法並非解決干擾問 題，而是依據自己的Traffic 需求，選擇合適的Configuration；另一作 法為Cluster 數個基地台，聯合進行同一Configuration 的策略，使彼此 間降低干擾。在Traffic Adaptation 中，各個基地台將會選擇最符合自 己traffic 樣貌的configuration，並不會顧及其他訊號發射源所帶來的干擾或者自身信號影響其他接收端的干擾，事實上便是沒有正面面對 干擾問題，而是盡量的滿足自己Traffic 需求，以求整體系統輸出達到 最大。而Cluster 通常會考慮path loss、SINR 或地理位置等，當作形 成Cluster 依據，考慮到彼此的干擾影響，互相配合，進而提高在UL 時的SINR，使得輸出因為SINR 的上升而提高，但各基地台需要共同 選擇對整個Cluster 最有利的configuration，卻可能因為此configuration 並不適合自己的Traffic 需求，而降低了系統Throughput 和部分基地台 的Throughput。本研究便是在期望在兩方法中間取得平衡，解決因為 不同Configuration 帶來的干擾問題，且不犧牲自己的Traffic 樣貌以委 屈於Cluster 的Configuration。因此本論文提出一個新的演算法，先以 類似Cluster 形式進行分組，並給予不同的編號，當在Cluster 成員因 受到干擾時，將會輪流選擇組內不同編號執行重新選擇configuration， 且選擇的方式根據分析過後的configuration map，迫使相鄰的基地台逐 步擇UL 較多的Configuration，以避免由基地台DL 造成的強大干擾， 以保護UL，並在一定時間後或者干擾情況解除時，各自重回Traffic Adaptation 狀態，使得在干擾情況嚴重時，避免干擾，又能在適當時 機自行選擇合適的Configuration。此外我們將討論Traffic Adaptation 和 Cluster 中原本在SINR 的差距，是否能在此間隙中取得適當的SINR 並換取足夠的throughput，便能達到保護UL 且提升整體DL 和UL 的 total throughput，並找尋在Total throughput 之外是否有其他優勢存在， 將會是本研究的重要課題。

In the LTE networks, there is a need to develop flexible schemes to adjust the UL and DL configuration of the UEs in order to meet the increasing demand for data due to the increasing use of multimedia. LTE Dynamic Time- Division duplexing (TDD) was created for this reason; it can ensure the base station selects a proper UL and DL pattern for different traffic demands in order to maximize the throughput of the small cell networks. However, Dynamic TDD’s advantage is limited in high density deployment scenarios due to the fact that the interference due to the number of base stations is very high.In the 3GPP specification, there are 7 different LTE TDD configurations; each configuration has a distinct UL / DL pattern. Each base station must select a configuration to use. Our research focuses on the case where the base station’s, femto cell’s, or other UE’s signal strength is different from that of other UEs. In this scenario, the signal strength from the UE is insufficient and the interference will cause the SINR to decrease. In the scenario of a UE being UL and another UE being DL, the SINR and throughput will be lower than the scenario where both are UL. There are two kinds of scheme for the base station. In the first scenario, traffic adaptation, each base station will choose the appropriate configuration for the traffic demand. In the second scenario,some base stations experiencing a high interference will cluster together, i.e. they will use the same configuration in order to mitigate the interference in UL-DL case; as a result, the SINR and throughput will increase in UL case. There is still a problem in the diverse traffic scenario as the configuration of the cluster is unable to match the traffic pattern of every base station in cluster. In this research, we discuss this scenario and balance the advantages of traffic adaptation with clustering. There are three parts to the algorithm. In the first part, the UEs will group as in the clustering and obtain the reconfiguration group tag, which will be used in reconfiguration. In the second part, if some base stations experience high interference, a high interference alert is raised and the group will reconfigured based on the reconfiguration group tag and the concession map, which helps mitigate the interference for DL. In the third part, each base station in algorithm will revert back to traffic adaptation when the high interference alert lifted.