The upcoming 5G standard has grabbed the industry spotlight. 3GPP divided 5G in two phases: Phase 1 (Rel-15) and Phase 2 (Rel-16). However, 3GPP agreed an accelerated plan for 5G deployment at the RAN plenary meeting held in March 2017 in Dubrovnik, Croatia. This means that 5G will come earlier than was originally envisaged.
As a multi-scenario integrated network, 5G provides the foundation for the ICT development. It will drive integration and innovation in terminal, wireless, network and service fields. The 5G network focus on user experience and will revolutionize the way people perceive, acquire and control information. Therefore, higher requirements are imposed on the transport network, in terms of bandwidth, connection density, latency, synchronization, cost and efficiency.
Huge Bandwidth Increase
In the ICT era, intelligent mobile terminals and cloud applications will promote the explosive growth of data. The 5G wireless base stations employ Massive MIMO, CoMP and high-order modulation to improve spectrum usage, and with the introduction of new spectrum, the bandwidth of a single base station can increase by tens of times. Take the 5G low-frequency base station (64 antennas, 16 streams, 100 MHz bandwidth and S111 configuration) for an example to calculate bandwidth requirements. As the average bandwidth required by a base station is over 2 Gbps and the peak bandwidth over 6 Gbps, six such base stations connected to a ring need to be configured with 20 Gbps bandwidth (peak bandwidth for one base station and average bandwidth for the remaining five base stations).
However, in practical deployments, the number of the antennas and streams, spectrum bandwidth, and the base station density may vary. Moreover, 5G high-frequency base stations also generate extra bandwidth demands. Therefore, in addition to providing bigger interface bandwidth, the transport network needs to support smooth bandwidth expansion (for example, link capacity expansion by bonding multiple links or wavelengths) so as to satisfy the ever-increasing bandwidth demands in the future.
The three main 5G usage cases of mMTC, uRLLC, and eMBB cover various applications such as mobile communications, ultra-high-definition videos, cloud office and games, VR/AR, smart wearing, smart home, smart city, industrial automation, automatic driving and highly reliable applications, which have differentiated latency demands. As suggested by NGMN, the one-way E2E latency in uRLLC scenario should be less than 1 ms. 3GPP sets uRLLC wireless air interface latency to 0.5 ms. Besides, there is latency in core network, and the backhaul network latency should be around 100 µs. These requirements present huge challenges to both 5G fronthaul and backhaul networks.
Highly Accurate Synchronization
The new 5G frame structure requires ±390 ns synchronization accuracy for the air interface. 5G inter-site CA and JT technologies requires ±130 ns synchronization accuracy (±5 ns for a single node of the transport network). Moreover, the high-precision positioning service with different positioning accuracy levels also has some special requirements for the synchronization.
In the 5G era, operators need a network architecture that supports unified deployment and operation. Coordination is needed across different control domains of RAN, backhaul network and core network to realize full service control and operation covering cloud, pipe and terminal layers. SDN/NFV technology is utilized to optimize data flow paths and make service source stay close to the service anchor point. This effectively shortens the network transport latency. Service-oriented APIs are constructed to meet differentiated service requirements and also make the service deployment more efficient. The network orchestration and management system is used to implement network slicing according to the needs of specific scenarios.
Meanwhile, SDN/NFV enables allocation of network resources to slices based on demand, where each network slice can have different application and control. This helps create an on-demand network architecture to support various application scenarios, opening up network capabilities and bringing new operation mode and potential profit.
In some 5G dense network scenarios, the cloud-RAN (C-RAN) can provide the flexibility in wireless resource management and function deployment to enable MEC service. It can also achieve decoupling of software and hardware to facilitate softwarization of cellular networks. However, if the CPRI interface is still used for 5G fronthauling, much higher bandwidth and much lower latency would be required. There is consensus that CPRI is not the right fit for 5G fronthaul. A new fronthaul interface is urgently required. Meanwhile, it is commonly believed that the BBU in 5G C-RAN will be split into two functional entities: centralized unit (CU) and distributed unit (DU). The CU mainly processes non-real-time protocols, and also supports the deployment of certain core networks functions in the network edge. The DU mainly handles physical layer functions, real-time HARQ flows and carrier aggregation. Some physical layer functions can also be moved downwards to RRU/AAU to significantly reduce the transmission bandwidth between RRU/AAU and DU and lower transmission costs. The bandwidth requirements between CU and DU are almost the same as those for the backhaul network. The interface between DU and RRU/AAU has not been standardized. Although solutions such as NGFI and eCPRI have already become available, it is estimated the standards will be finalized when the 5G NR protocol stack becomes mature and stable.
The evolution towards 5G will require a restructuring of operator networks. The traditional centralized core network is evolving towards a virtualized distributed core that places the user plane close to users. The user plane functions of 5G core network will be moved to the network edge since the deployment of service anchor points is also distributed. The backhaul network not only needs to support dynamic anchor point selection, but also satisfies the inter-cloud traffic requirements after the network is re-architected and cloudified.
Compared with conventional mobile communications systems, 5G puts more stringent requirements on the transport network. As the cornerstone of the 5G network, the transport network needs to introduce new transport interfaces, technologies and network control capabilities, adapt to diverse network architectures, meet demands for bandwidth, differentiated latency, highly accurate synchronization, network slicing and enhanced network openness and coordination, facilitating the continuous network evolution.
5G transport, network slicing, latency, C-RAN, ubiquitous connections