IPX ECOSYSTEM AND RELATED WORK

The mesh of interconnected IPX-Providers (IPX-Ps), and their Service Providers (SP) customers form the IPX Ecosystem (Figure 1). IPX-Ps are third party interconnection providers to Service Providers (SPs) such as MNOs, IoT providers or cloud providers. They provide support for global data roaming and a variety of emerging services, such as IoT, VoLTE or Rich Communication Services (RCS).

IPX-Ps peer (via private interconnects or public peering) with other IPX-Ps to extend their footprint worldwide forming the IPX Network – a private network, separate from the public Internet, that meshes together the infrastructures of the IPX-Ps. It enables the transport of global roaming data between networks, with interoperability of different implementations and standards.

SPs require a single connection and agreement with one IPX-P in order to connect to the IPX Network, and interconnect with partner SPs world-wide. 1 For instance, to enable data roaming, two MNOs must both have an agreement with an IPX-P in order to interconnect. For redundancy, a SP could establish physical connections to more than one IPX-P. Depending on the footprint of the IPX-P’s infrastructure, SPs select one or more Point of Presences (PoPs) of the IPX-P to connect. In an earlier publication [18], we described the IPX ecosystem and how it enables data roaming.

由相互连接的IP交换提供商（IPX-Ps）及其服务提供商（SP）客户构成的网络共同形成了IPX生态系统（见图1）。IPX-Ps是为服务提供商（SPs）（如移动网络运营商（MNOs）、物联网（IoT）提供商或云提供商）提供互联服务的第三方提供商。它们为全球数据漫游以及包括物联网、LTE语音（VoLTE）或富通信服务（RCS）在内的多种新兴服务提供支持。

IPX-Ps通过（私有互联或公共对等方式）与其他IPX-Ps进行对等互联，以扩展其全球覆盖范围，从而形成IPX网络——这是一个独立于公共互联网的私有网络，它将各IPX-Ps的基础设施融合为一个整体网络。该网络使得全球漫游数据能够在不同网络间传输，并确保不同实现和标准之间的互操作性。

服务提供商（SPs）仅需与一家IPX-P建立单一连接并签订一份协议，即可接入IPX网络，并与全球范围内的合作伙伴SPs实现互联。例如，为了实现数据漫游，两个MNOs都必须与某家IPX-P签订协议才能实现互联。为了实现冗余，服务提供商（SP）可以与不止一家IPX-P建立物理连接。根据IPX-P基础设施的覆盖范围，服务提供商（SPs）会选择一个或多个IPX-P的接入点（PoPs）进行连接。在我们先前发表的文献[18]中，我们描述了IPX生态系统及其如何实现数据漫游。

Related Work

IPX model and network. The IPX model was first proposed by the GSMA in 2007 to replace the traditional, bilateral-agreement model for international roaming [5]. Despite the continuous evolution IPX-Ps and related parties [7, 33, 34] the topic has received little attention from the research community. Takaaki [23] provides an early survey of IPX and its technical requirements. Recently, in [18] we analyzed the IPX network and reported it is composed of 29 active IPX-Ps peering via three major peering exchange points, and interconnecting about 800 MNOs worldwide. [18] also showcases the radio signaling infrastructure of a commercial IPX-P and reports high-level trends. This paper presents the first in-depth analysis of a commercial IPX-P, providing implications for its operations and for its customers.

IPX模型与网络。IPX模型最初由GSMA于2007年提出，旨在取代传统的国际漫游双边协议模式[5]。尽管IPX-Ps及相关方在持续演进[7, 33, 34]，但该主题在研究界受到的关注甚少。Takaaki [23]对IPX及其技术要求进行了早期综述。近期，我们在[18]中分析了IPX网络，并报告其由29家活跃的IPX-Ps组成，这些IPX-Ps通过三个主要的对等交换点进行互联，连接了全球约800家移动网络运营商（MNOs）。[18]还展示了一家商用IPX-P的无线信令基础设施，并报告了其宏观趋势。本文首次对一家商用IPX-P进行了深入分析，为其运营及其客户带来了启示。

Roaming. Few studies have been conducted on roaming, possibly because its complex ecosystem and many involved parties bring about high costs and efforts for cooperation. Vallina et al. [31] studied national roaming between MNOs in France, and Michelinakis et al. [22] focused on international roaming between two operators in Europe. More recently, Mandalari et al. [21] covered international roaming more extensively, diving into the traffic among 16 MNOs in 6 different countries. Differently, this paper focuses on the operational insights of a commercial IPX-P, with customers in 19 countries which receive inbound roamers from 215 countries and whose subscribers travel to 210 countries.

漫游。关于漫游的研究较少，这可能是因为其复杂的生态系统和众多参与方导致了高昂的合作成本和精力投入。Vallina等人[31]研究了法国MNOs之间的国内漫游，Michelinakis等人[22]则聚焦于欧洲两家运营商之间的国际漫游。最近，Mandalari等人[21]更广泛地研究了国际漫游，深入分析了6个不同国家中16家MNOs之间的流量。不同的是，本文专注于一家商用IPX-P的运营洞察，该IPX-P的客户遍布19个国家，接收来自215个国家的入访漫游用户，其签约用户则漫游至210个国家。

IoT/M2M. There is a large body of work on IoT/M2M traffic. In particular, numerous research shows that IoT devices generate traffic with significantly different patterns from human-driven mobile devices in the cellular networks [8, 15, 16, 27]. Based on such observations, Markus et al. [17] tried to fabricate M2M traffic models, and the other researchers design future systems that can efficiently handle these traffic [2, 24]. Lutu et al. [19] analyze IoT traffic from the point of view of an IoT provider and MNOs, typically customers of IPX-Ps. While this study does not focus on IoT, we complement previous work by analyzing the M2M service provided by a commercial IPX-P and provide operational insights. Our results are in line with recent work [14] that characterizes IoT signaling traffic from a network operator’s point of view for the establishment of data connections at device level.

物联网（IoT）/ 机器对机器（M2M）通信。关于物联网/M2M流量已有大量研究工作。特别是，众多研究表明，在蜂窝网络中，物联网设备产生的流量模式与人类用户驱动的移动设备显著不同[8, 15, 16, 27]。基于这些观察，Markus等人[17]尝试构建了M2M流量模型，其他研究人员则设计了能够高效处理这些流量的未来系统[2, 24]。Lutu等人[19]从物联网提供商和MNOs（通常是IPX-Ps的客户）的视角分析了物联网流量。虽然本研究并非专门针对物联网，但我们通过分析一家商用IPX-P提供的M2M服务并提供运营洞察，对以往工作进行了补充。我们的结果与近期研究[14]一致，该研究从网络运营商的视角描述了用于在设备层面建立数据连接的物联网信令流量特征。