5G architecture innovation is obvious and the bearer network should fully support it

5G will become one of the hot technologies in the information and communication field in the next five years. It will focus on application scenarios such as mobile Internet and mobile Internet of Things, and present characteristics such as ultra-large capacity, ultra-high performance, high reliability and low cost. At the same time, it will develop in coordination with artificial intelligence, information center networks, etc.

Judging from the key parameters currently set, for different application scenarios, the access peak rate experienced by 5G users will reach tens of Gbit/s, which is a hundred times or more higher than that of 4G. The end-to-end delay requirement is at the ms level or lower, and the synchronization accuracy needs to reach the 100ns level. In terms of networking architecture and technology, 5G needs to introduce new technologies such as software-defined networking (SDN), network function virtualization (NFV), and network slicing. These new features of 5G have put forward many challenging requirements for mobile fronthaul, mobile backhaul and other bearer networks, and also promoted a new round of development opportunities for the transport network for 5G bearer.

[[199914]]

Base station functions are redistributed, and fronthaul presents networking requirements

Considering that 5G has much higher requirements in terms of capacity, latency, synchronization accuracy, etc. than 4G, the original C-RAN architecture and bearer technology based on indoor baseband processing units (BBU) and remote radio units (RRU) will face challenges, and the network element functions of the 5G wireless access network will be re-divided and strive to achieve flexible settings on demand. At present, 3GPP divides 5G base stations into central units (CU) and distributed units (DU) according to logical functions, and can nest and divide intermediate CU or DU units between CU and DU according to application requirements. In this way, there will be a networking bearer demand between DU and CU, and different bearer technologies can be selected according to the functional division of DU and CU, deployment methods and service types, such as the emergence of structures such as primary fronthaul and secondary fronthaul, but it does not exclude the coexistence of structures similar to 4G fronthaul.

Traffic attribution has changed significantly, and the need for backhaul coordination is obvious

Compared with 4G networks, 5G network architecture will be flatter, and some existing and newly added functions (such as mobile edge computing, etc.) will be moved down to the aggregation layer or lower. The original typical traffic pattern based on north-south aggregation will change significantly, and the east-west traffic interaction between different 5G network elements will increase significantly, such as gNB and gNB, CU and CU, etc., which puts forward new requirements for the east-west collaborative traffic scheduling of backhaul networks and even fronthaul networks. In addition, with the downward migration of some core functions of 5G, functions based on the third layer (IP layer) will also be moved down to the aggregation layer, and some new routing technologies (such as SR, etc.) will be needed to improve routing efficiency. Therefore, how to effectively segment and efficiently coordinate the networking functions of the transport network and the IP network is also one of the problems that need to be solved.

5G network architecture is more complex

The 5G network architecture separates control and bearer (forwarding) functions based on SDN, and proposes the concept of logical network slicing to flexibly support cloud services. In order to adapt to the functional requirements of 5G networking, the transport network needs to simultaneously introduce an SDN-based optical transport network architecture, and combine the diversified bearer requirements of 5G, so that the fronthaul, backhaul and core network transmission networks can work together on demand. In addition, the transport network should also gradually introduce logical network slicing technology and support corresponding functions from different dimensions such as the application plane, control plane and transport plane. For example, under the SDN network architecture, multiple differentiated virtual topologies are constructed and abstracted on the same physical topology based on different ODUk time slots, different wavelengths, different packet pipes, etc., and 5G services with specific functions and performance requirements are carried on demand.

Standardization research is accelerating, but the transmission format is yet to be selected

At present, international standardization organizations and groups such as ITU-T, 3GPP, IEEE, and CPRI have all started standardization work related to 5G bearer. For example, ITU-T SG15 established two new research topics at the plenary meeting in June 2017. 3GPP is promoting 5G standardization and considering bearer requirements. IEEE1914 is conducting research on the new generation of fronthaul interfaces and adaptation. CPRI is standardizing new fronthaul interfaces (eCPRI). In addition, bearer technologies of different standards are also evolving at an accelerated pace. Considering the requirements of ultra-large capacity, ultra-low latency, ultra-high synchronization accuracy, and flexible networking, in addition to universal physical layer technologies such as fiber direct drive and wavelength division multiplexing technology, networking technologies based on 1.5 layers/2 layers are currently the most competitive, including FlexO+ODUflex based on OTN and its simplified standards combined with some packet functions, and FlexE based on packetization mode. The former is relatively mature, but the functions and structure may need to be further simplified. The latter needs to evolve from interface to networking technology, which still needs the full promotion of the industry. The specific choice of which standard to adopt still depends on technological competition and industry-based selection.

Author: Wu Bingbing and Zhao Wenyu, Institute of Technology and Standards, China Academy of Information and Communications Technology