Any to Any MPLS for Mobile Backhaul

With the wide application of mobile data services and the emergence of new services, especially mobile video services, mobile network traffic has increased dramatically.
Major operators worldwide have been upgrading their wireless telecom networks over the past decade from 2G to 3G to LTE, to meet the demand of higher bandwidths. While continuous network upgrades are providing users with improved network performance and QoS, they are also creating many challenges in mobile backhaul networks.

To accommodate LTE, the next-generation mobile backhaul network must meet the following requirements:
●   Smart network architecture. The evolution of mobile networks toward 4G makes a different LTE network architecture from 2G/3G one. Specifically, part of the control and routing functions previously implemented by the BSC/RNC are integrated into the eNodeB to make a flatter network capable of smart routing. Higher network bandwidths will also be provided with both uplink and downlink bit rate at above 150 Mbps. 1 GE and even 10 GE access devices will become common.
●    Super-large scale with densely-deployed LTE eNodeBs. Either the rapid development of LTE technologies, urgent demand for higher user bandwidths, or exponentially growing number of eNodeBs requires operators to deploy more access nodes on the transport network to handle huge accesses to eNodeBs and carry large service traffic.
●    Smooth and seamless service migration to the LTE network. A well-established LTE network requires a long-term process and will not come overnight. Usually, there is only a small number of users at the early stage, and the LTE EPC core network is deployed in one region and shared by users in this and other regions through transport networks. However, the increasing number of users and rising demand for user capacity urge operators to consider deploying the EPC to be closer to users, which means from a centralized EPC to distributed EPCs in multiple regions.
●    Low transmission delays. LTE services require lower delay than legacy 2G and 3G services. According to 3GPP2 and NGMN, the end-to-end delay of S1 interface in LTE should be less than 25 ms, with an ideal value of less than 5 ms, and the end-to-end delay of  X2 interface in LTE should be between 50 ms and 100 ms, with an ideal value of less than 10 ms.
●    Compatible and converged transport capability. Many operators have to deploy and operate mobile networks with different network systems during LTE evolution. Therefore, a next-generation mobile backhaul network must support multiple types of existing transport links, such as copper, fiber, and radio (mainly MW). Not only should it provide TDM/ATM access for all legacy 2G/3G services to preserve investments in the existing backhaul networks, it must also be highly scalable for future use.
●    Intelligent O&M In the traditional "black box" O&M mode, network performance is separated from QoS. The network management system focuses on network performance and is not aware of QoS and how network resources are allocated. An evolved backhaul will allow the network to carry more diversified 3G and LTE services and make network O&M more complex as well. Therefore, an intelligent O&M system with clear service paths, visual service resources, and abundant real-time and smart QoS monitoring methods will be indispensable to a future mobile backhaul network.

After years of studying the mobile backhaul network, ZTE proposed the Any to Any MPLS solution for the next-generation mobile backhaul network. This solution relies on a seamless MPLS architecture to support all applications on the existing mobile backhaul network. It has the following features:
●  Any media. Supports all access media including fiber, cable, and microwave.
●  Any topology. Applies to all network topologies, such as a circular, horseshoe-shaped, or dual-uplink network.
●  Any service. Supports all 2G, 3G, and 4G services; satisfies FMC evolution requirements; and provides access in special scenarios, such as triple play users in the fixed network, wholesale, and carrier of carrier.
● Any location. Allows users to access the network from any location and communicate with other users in any location.
●  Any scale. Carries services on all small, medium, and large networks.

●  Any environment. Applies in all harsh environments such as thunderstorms, lightning, strong wind, sandstorms, and snowstorms. Unlike previous solutions, Any to Any MPLS solution (Fig.1) separates the transport plane from the service plane. It uses the BGP-LU (based on FRC3107) function so that the network can process services only on the end nodes at the network border and make the rest of the network a transport tunnel without control over and exchange with the service plane. It is based on the following ideas:
●  Routing domain division. The network is divided into multiple small routing domains to accelerate route convergence and reduce the workload of devices within each routing domain.
●  Layered route reflector (RR) design. On the service layer and transport layer, the existing devices can be used as inline RRs or special devices can be used as external RRs to reduce the number of BGP sessions in the network and provide step-by-step reflection and an end-to-end VPN/PW service. When a service accesses the network, configurations need to be done only on two end nodes at the border.
●  Hierarchical LSP design. An LDP-based LSP is configured for each routing domain for internal use, and services within the routing domain are distributed directly through IGP+LDP. Services involving multiple routing domains are distributed through hierarchical LSPs. Specifically, BGP-LU and the corresponding next-hop-self are modified to learn the loopback address of the remote end.
●  Route filtering. The UPE receives many BGP VPN routes advertised by the SPE and other remote nodes. It also receives labeled BGP unicast IPv4 routes as well. Therefore, the SPE and NPE can be configured with the prefix filtering function to filter unnecessary BGP VPN routes and labeled BGP IPv4 routes in order to avoid needless route diffusion and lower the routing workload of the UPE.

Any to Any MPLS solution has the following advantages:
●    Complete decoupling of the transport plane and service plane. The previous node paste, connection and label switching among intermediate nodes are not used any more. Service nodes at the network edge are configured to provide end-to-end services with the rest of the network serving as a transport tunnel.
●    Flexible service deployment. The Any to Any MPLS solution uses prefix filtering to eradicate the defects of the previous solutions and carry multiple services through a unified technology.
●    Rapid convergence and protection. This solution uses routing domain division and rides on IGP FC, LDP FRR, BGP-PIC, and other technologies to provide rapid switchover and protection within 50 ms.
●    High scalability. Routing domain division is used to logically divide a large, single MPLS-based network into smaller isolated networks. This is more suitable for scale-out scenarios and theoretically can support super-large networks in the future.
●    Easy O&M. This solution supports a full range of O&M tools and specialized ZXDNA traffic analysis tools, to provide network planning, design, commissioning, diagnosis, troubleshooting, and optimization, achieve one-click distribution, locate faults rapidly, identify services intelligently, and analyze and control traffic automatically.