The increasing reliance on wireless connectivity comes with a hefty cable infrastructure with many realms of technology converging to provide seamless wireless service to laptops and handheld devices. These technologies range from the Common Public Radio Interface (CPRI) fiber backhaul for 5G wireless communications base transceiver stations (BTS) to the innovative coaxial connectors used for antenna feeds. A number of proposed backhaul, fronthaul and midhaul technologies can be used to support the high capacity and latency demands, including satellite, millimeter-wave, microwave, copper and fiber backhaul. The following information focuses more on the wired interconnections that will make 5G possible.
The Backhaul Challenge
The general idea of the 5G backhaul network is to send data from the distributed next-generation NodeBs (gNBs) or 5G New Radio (NR) base stations or cell sites to the 5G core network (5GC). With this in mind, the goal of 5G is to build upon previous generations of cellular networks without replacing them outright, leaving intact LTE Evolved Packet System (EPS) architectures based on orthogonal frequency-division multiplex (OFDM) modulation with Ethernet physical interfaces and more legacy 2G/3G GSM architectures based on time-division multiplexing (TDM) T1 physical interfaces. The aim to build upon previous cellular technology was realized in late 2017 with 3GPP’s Release 15, wherein non-standalone and standalone 5G configurations were specified.
With regard to backhaul, the key performance indicators (KPIs) for 5G data improvements require much more research and development (R&D) and capital expenditures. They are as follows:3
1,000 times capacity compared with 2010 (1–10 Gbps/sector)
Decrease in latency (~1 millisecond E2E)
99.999 percent availability (zero perceived downtime)
100 percent coverage (ultra-dense network (UDN) to connect 7 trillion user equipments (UEs).
Saving 90 percent of energy per service provided
Reducing average service creation time cycle from 90 hours to 90 minutes
The wireless solutions proposed to solve these prominent 5G concerns are heavily investigated (e.g., carrier aggregation [CA], massive multiple-input multiple-output [MIMO] antennas, high-order quadrature amplitude modulation (QAM) schemes, millimeter-wave technologies, additional spectrum usage and small cell deployment). The current cellular backhaul will not readily support the capacity, latency and reach requirements (see Table 1).
The 5G network architecture attempts to solve some of these major data challenges with UDNs of small cells and macrocells, mobile edge computing, centralized cloud processing, radio access network (RAN) virtualization and the decoupling of the user and control plane. Additional processing necessarily occurs with the UDN as interference and cellular hand-offs grow rapidly in complexity. Solutions such as intercell interference coordination (ICIC) schemes and coordinated multipoint processing (CoMP) come with a hefty power consumption. The proposed centralized RAN (C-RAN) 5G network aims to split the processing, allowing for smarter power usage while enabling user mobility. The next generation RAN (NG-RAN) architecture requires not just a backhaul portion of the network, but also a fronthaul and midhaul in many cases. The current backhaul infrastructure simply cannot support the innovations and capacity restraints. This introduces a realm of possibility between wired solutions and millimeter-wave as well as microwave wireless backhaul opportunities.
Proposed 5G Network Architecture
The 3GPP NR architecture involves a gNB that consists of a central unit (gNB-CU) and one or more distributed units (gNB-DU). The baseband unit (BBU) in 4G iterations is now split into the remote radio unit (RRU), CU and DU. The CU processes non-real time protocols and services while the DU processes PHY level protocols and real-time services.1 The separation of radio and processing allows for multiple tiers of service, separating time-sensitive services for low-latency communications and compute-intensive applications as well as the refrigeration and cooling that comes with it. From a hardware perspective, the RRUs include an RF front-end (RFE) while the baseband processing occurs at the DU and CU. Although most 3GPP projects consider the CU and DU split, some include more specific extensions dividing the DU into the RRU and DU. The RRU and DU split typically introduces Common Public Radio Interface (CPRI) or enhanced CPRI (eCPRI) as the communications interface between the two. In 2G/3G, T1 was the interface relied upon for backhaul. In 4G LTE, the X2 and S1 were the interfaces between the RAN nodes and core network (serving gateway [SGW] and PDN gateway [PGW]). In 5G NR, the logical interfaces are XN, NG, E1 and F1. Figure 1 offers an overview of the 5G network. The general interface connections can be defined as follows:
XN: Interface between base stations (user plane interface).
NG: Interface between base stations and the core (control plane interface).
F1: Interface between CU and DU.
E1: interface between CU-user plane (CU-UP) and CU-control plane (CU-CP).
The International Telecommunications Union (ITU) GSTR-TN5G 2018 Technical Report defines fronthaul, midhaul and backhaul as follows:
Fronthaul: the network between RRUs and DUs. CPRI or newer eCPRI interfaces.
Midhaul: the network between the CU and DU (F1) or the network between gNBs (Xn).
Backhaul: the network between the CU and the 5GC. NG interface.
Table 1 lists the capacity, latency and network reach requirements for 5G interfaces and transport networks. These requirements seem to make evident the need for more fiber deployments to handle the capacity and reach necessary for 5G.
CPRI: Fiber is a critical technology for 5G fronthaul, midhaul and backhaul. The general consensus involves a significant increase in the use of fiber and millimeter-wave technologies for backhaul over sub-6 GHz, microwave and satellite backhaul methods.2 Several proposed transport network infrastructures rely on fiber when using standard digitized formats such as CPRI and eCPRI from the RRU to the DU, and from the DU to the CU. From distributed antenna system (DAS) networks to current cellular base stations, the CPRI protocol is currently being used to connect remote radio heads (RRHs) to the baseband units (BBUs). Benefits include reach (over tens of kilometers), low latency (<5 microseconds), a maximum bit error rate (BER) of 10-12, support of high-order modulation schemes (64QAM) and compatibility with low-cost optical transceivers (e.g., SFP/SFP+). The main benefit of CPRI is the functional split between radio processing and digital processing, allowing the RRHs to be small and cost-effective. As such, CPRI can continue to uphold its benefits for 5G centralized RAN (C-RAN) architecture.
The fiber infrastructure for dense small cell networks can be a massive undertaking. This is especially true for last-mile installations from the small cell to the local DU aggregation point (often a local macrocell). Alternative technologies such as next-generation fronthaul interface (NGFI) and 5G-XHaul aim to eliminate the stringent requirements of CPRI by using lower-cost and flexible interfaces such as Ethernet and millimeter-wave links in cases where the size of the capital expense (capex) is a major barrier.
eCPRI: The CPRI protocol will not suffice for the data rate demands that come with massive multiple-input multiple-output (MIMO) communications and the use of large spectral bandwidths (hundreds of megahertz). The enhanced CPRI (eCPRI) protocol is meant to scale with the bandwidth more effectively than traditional CPRI transport. Moreover, eCPRI allows for interworking with CPRI over an Ethernet interface. This, in essence, allows for the coexistence of new and legacy equipment in the same Ethernet-based fronthaul networks.
Small Cells with Integrated Access and Backhaul (Mesh)
Some small cell backhaul solutions propose a hybrid wireless/fiber backbone where a small packet of access points (APs) support local traffic through a virtual network. As such, wireless backhaul through mesh networking can be used until a fiber attachment point.
Fiber-Wireless (FiWi) Point-to-Multipoint (PtMP)
Another solution involves a hybrid millimeter-wave/fiber backhaul to the aggregation point. The main difference between this and the previous solution is the lack of mesh connectivity. A millimeter-wave pencil beam is sent from a fiber-connected RRH to various RRHs via a line-of-sight (LoS) link through an optical beamforming network (OBFN). The remotely connected RRHs are fiber-connected to RUs that service local UEs. This type of backhaul relies on the eCPRI connection from the RU to the RRHs as well as the PtMP access from the massive MIMO V-band installation.
The NG-RAN infrastructure relies heavily upon deep fiber and last-mile fiber deployments as well as hybrid wireless solutions to support the increase in capacity and latency listed in 5G’s KPIs. Much of the current cellular infrastructure is still relied upon for aggregation while massive small cell deployment is occurring in cities around the globe. The UDN comes with a major challenge of backhaul with competing wireless and wireline solutions. Although there is certainly no one-size-fits-all solution, digital radio over fiber (DRoF) and analog radio over fiber (ARoF) are extremely important for providing a reliable transport network over long reaches.
Oliva, Antonio De La, et al. “An Overview of the CPRI Specification and Its Application to C-RAN-Based LTE Scenarios.” IEEE Communications magazine, vol. 54, no. 2, 2016, pp. 152–159., doi:10.1109/mcom.2016.7402275.
Jaber, Mona, et al. “5G Backhaul Challenges and Emerging Research Directions: A Survey.” IEEE Access, vol. 4, 2016, pp. 1743–1766., doi:10.1109/access.2016.2556011.
Ahamed, Md Maruf, and Saleh Faruque. “5G Backhaul: Requirements, Challenges, and Emerging Technologies.” Broadband Communications Networks – Recent Advances and Lessons from Practice, 2018, doi:10.5772/intechopen.78615.
Paul Hospodar is product manager at L-com.