基於網路系統壽命最佳化的D2D協同合作重傳機制

Abstract

隨著智慧型手機的普及，使用者對行動上網的網速需求提高，長期演進技術(Long Term Evolution, LTE)與長期演進技術升級版(LTE-Advanced)因而發展，LTE為全IP的扁平化網路，下行速度高達100Mbps，因此使用者使用手機下載資料的情況越趨頻繁。裝置間通訊(Device-to-Device Communication, D2D)是LTE、LTE-A的重要議題，其好處是使用者設備(User Equipment, UE)之間可以直接通訊而不需要利用基地台(eNB)轉傳、可以降低資料遺失率並且增加LTE上行頻寬的使用率。資料遺失在資料傳輸的過程中是很常發生的現象，然而現行針對D2D協作重傳的文獻數量很少，且具有效率過低、無法確保能完全重傳給所有需要的使用者或是D2D簇(D2D Cluster)的壽命太短等缺點。在本文章中，我們發展了一個D2D協同合作重傳機制，目的是利用D2D 通訊(D2D communication)的優點替eNB協作重傳，我們設計了一個尋找重傳組的演算法，從正確接收封包的裝置(ACKed UEs)中選取具有高覆蓋率、低重複數、高剩餘電量者當作重傳者，找到的重傳組除了能完整涵蓋所有的未成功接收封包裝置(NACKed UEs)，此演算法的執行時間短，找出的重傳者彼此之間不相交(Disjoint)程度高，並能延長D2D簇的網路壽命。每當重傳完一個封包後，eNB就會計算ACKed UEs的剩餘電量，並且依據預先設定的剩餘電量的臨界值調整ACKed UEs的重傳功率，如此一來能提升ACKed UEs的壽命，並最大化D2D簇網路壽命。本論文使用C language撰寫實驗模擬，實驗結果顯示我們的方法相較於其他論文，具有較短的執行時間與較長的網路壽命。

With the popularity of smart phones, Long Term Evolution (LTE) has been developed. Since the downlink speed of LTE can be up to 100Mbps, increasing users download data by cell phones. Device-to-Device (D2D) communication is an important issue in LTE. The benefits of D2D communications are: User Equipments (UEs) can communicate directly, the system has a lower packet loss rate (PLR) and etc. We take advantage of D2D communications to do the retransmission, which is still a challenging topic. In this thesis, we develop a new D2D cooperative retransmission scheme to do the retransmission. We design an algorithm to find an optimal retransmission set which selects the ACKed UEs with a higher degree, lower repetition, higher remaining battery capacity as the retransmitters to retransmit the data to NACKed UEs, and the set of selected ACKed UEs is R, which can cover all NACKed UEs. The execution time of the algorithm is short and it can prolong the network lifetime. After data retransmission, eNB calculates the remaining battery capacity of the ACKed UEs, if their power level is lower than a predefined threshold (T), their retransmission power will be reduced to prolong the node lifetime. The simulation results show that our approach indeed has shorter execution time and a longer network lifetime.