Design Overview

To mitigate the handover problem in mobile satellite network, we propose a novel scheme for reducing the handover latency by making full use of the predictable satellite trajectory and spatial distribution of satellites.

为缓解卫星移动网络中的切换问题，我们提出了一种新方案。该方案充分利用卫星轨迹的可预测性以及卫星的空间分布特性，以降低切换延迟。

We divide the overall handover procedure in mobile satellite network into three parts: UE to the Radio Access Network (RAN), RAN to RAN, and RAN to the core network. We conducted preliminary experiments in to investigate the latency brought by these different steps. As shown in Fig. 3, we find that the time overhead consumed by the transmission from RAN to the core network dominates the overall handover latency. The reason is that this process consists of passing through multiple ISLs and the satellite-ground link. Therefore, the main aim of this work is to mitigate or even eliminate the interaction between RAN and the core network, thus considerably reducing the handover delay in mobile satellite network.

我们将卫星 移动网络中的整体切换流程划分为三个部分：用户终端（UE）到无线接入网（RAN）的切换、无线接入网之间的切换（RAN to RAN），以及无线接入网到核心网（Core Network）的切换。

我们进行了初步实验，以探究这些不同步骤所带来的延迟。如图3所示，我们发现 从无线接入网到核心网的传输所消耗的时间开销在整体切换延迟中占主导地位。其原因在于，该过程包含了经过多条星间链路（ISLs）及星地链路的传输。

因此，本研究的 主要目标是减少乃至消除无线接入网与核心网之间的交互，从而显著降低卫星移动网络中的切换延迟。

In this paper, we have redesigned the handover signaling procedure to avoid interactions between the core network and RAN. However, this design introduces two new challenges. The first challenge is related to the synchronization problem between the core network and RAN since there exists no control signalling interaction between them, yet in the standard handover procedure of mobile satellite network there exists much interaction between the core network and RAN to ensure synchronization. The second challenge is the heavy computation overhead on the core network which may paralyze the core network since much trajectory prediction should be performed at the core network side.

在本文中，我们重新设计了切换信令流程，以避免核心网与无线接入网之间的交互。然而，这一设计引入了两个新挑战:

第一个挑战与 核心网和无线接入网之间的同步问题有关，因为它们之间没有控制信令的交互 ；而在标准的卫星移动网络切换流程中，核心网与无线接入网之间存在大量交互以确保同步。

第二个挑战是 核心网侧沉重的计算开销 ，由于大量的轨迹预测需要在核心网侧执行，这可能会导致核心网瘫痪。

To deal with the first challenge, we propose a fine-grained synchronized algorithm. Specifically, according to predictable trajectory of satellites and weather information, we predict the UE’s access satellites at two time points with a fixed interval without interaction with RAN, which then are utilized to determine whether the handover happens. For instance, we employ the simple yet effective binary search to achieve accurate prediction of the handover triggering time.

We address the second challenge by leveraging the satellite access strategy and the unique spatial distribution of LEO satellites to significantly reduce the number of UEs and satellites required for prediction and hence relieve the computational pressure for the core network.

为应对第一个挑战，我们提出了一种细粒度的同步算法。具体而言，我们根据可预测的卫星轨迹和天气信息，在无需与无线接入网交互的情况下，预测用户终端在两个具有固定时间间隔的时间点所接入的卫星，并利用该预测结果来判断切换是否发生。例如，我们采用简单而有效的二分查找法来实现对切换触发时间的精确预测。

为应对第二个挑战，我们利用卫星的接入策略以及LEO卫星独特的空间分布特性，来显著减少需要进行预测的用户终端和卫星的数量，从而减轻核心网的计算压力。

Furthermore, we have introduced several extra designs to reduce the handover latency and improve the system robustness. For example, we add a simple constraint (i.e., selecting the access satellite in similar travelling direction with the previous connected satellite) in the satellite access scheme. Later analysis and experiments will demonstrate this easily achievable aim can considerably reduce the handover latency. Simultaneously, we investigate the impact of inaccurate prediction on the performance of the proposed handover scheme. To address its two main causes—user mobility and deviation in satellite trajectory prediction, we discuss and design corresponding strategies to deal with potential failures.

此外，我们还引入了一些额外设计来进一步降低切换延迟并提升系统鲁棒性。例如，我们在卫星接入方案中增加了一个简单的约束（即，选择与前一个连接卫星行进方向相似的卫星作为接入卫星）。后续的分析和实验将证明，这个易于实现的目标可以显著降低切换延迟。同时，我们研究了预测不准确对所提出切换方案性能的影响。针对其两个主要成因——用户移动性和卫星轨迹预测的偏差，我们讨论并设计了相应的策略来处理潜在的失败情况。