一個在S1 介面支援下使用e-MBMS 服務的新換手問題

Abstract

在4G LTE下，e-MBMS是未來提供直播服務的趨勢。為了要達到3GPP所定義的安全和收費要求，用來保護e-MBMS服務的金鑰組是需要做更新的。然而使用者的移動速度隨著科技的演進變得更加快速，也就是說UE需要經過更多次的換手才能繼續使用MBMS的服務。在這個研究中，我們指出一個在使用S1介面進行換手時，MBMS的服務產生換手失敗的問題。這個問題發生在換手時，若是兩基地台間沒有存在X2介面，在目前所連接基地台內的緩存區的資料將會遺失。若是金鑰組的更新資訊在換手時仍在緩存區內，那麼該用戶將得不到最新的金鑰組，也就是說該用戶無法解析此後用新的金鑰組所加密的資料。在該UE向BM-SC重新認證成功以前，MBMS的資料將全部無法解析，因而對於用戶體驗帶來極大的影響。並且這些重新認證的訊息交換，也將帶給核心網路一些額外的負擔。在這個研究中，我們提出了一個研究模型來分析這個問題，並且提出了三個可能的解決方法來減少對於使用者與核心網路的影響。最後我們提供詳細對於這個問題的詳細分析。這個研究中的模型以及討論將使電信營運商了解我們所提出的三種方法的優缺點，並且使其透過此研究中理論的分析來改變其網路佈置來減少上述問題的發生。此外，我們的研究模型經由大量的ns2模擬得到驗證。

evolved-Multimedia Broadcast Multicast Service (e-MBMS) is the future trend of delivering live streaming to mobile equipments in 4G Long Term Evaluation (LTE) network. In order to achieve security and charging requirements, e-MBMS key set update is needed. However, the moving speed of User Equipments (UEs) becomes higher in the future which means that an UE has to perform more handovers to continue its MBMS session. In this paper, we point out a new handover failure problem caused by S1 supported handover in MBMS session. Once a handover is performed without X2 interface support (also called S1-based handover), the buffered data in source eNodeB would be lost. If rekeying data happens to be in the lost data, the UE cannot get new keys to access ongoing communication encrypted by the new keys. This problem causes huge negative impacts to user experience since the UE cannot decrypt MBMS contents until the re-authentication procedure involved with Broadcast/Multicast Service Center (BM-SC) is done. In addition, this problem also introduces extra signaling overhead caused by the re-authentication procedure for core network. In this paper, we present a math model to analyze this problem, and propose three possible solutions to reduce the negative impacts. We then offer detailed discussions over the model. The model and discussions give a clue for operators to understand the suitable conditions of those three solutions, and is also a theoretical instruction for operators to reduce handover failure by adjust their network deployment settings. Furthermore, our analytical model is validated by extensive simulations by using ns2.