Exploring new centralized RAN and fronthaul opportunities

The term CRAN has been around for many years, with different meanings including Centralized RAN, Coordinated RAN and Cloud RAN. In this blog, the term CRAN refers to centralized RAN, although we will also touch on coordination and cloud. Centralized RAN (CRAN) is an alternative to the most common deployment option, de-centralized RAN (DRAN). In DRAN, all processing for each antenna site is done locally at the antenna site. In CRAN on the other hand, a large part of the processing is done at CRAN hub for multiple antenna sites. There are many different split options for the RAN protocol stack. In this blog, we focus on the so-called lower layer split (LLS) in Layer 1 where Layer 3, Layer 2 and part of Layer 1 are processed in the CRAN hub.

In DRAN, the transport segment from antenna site to core network is referred to as the backhaul; in CRAN, the interface between the antenna site and the CRAN data center is referred to as the fronthaul. Generally speaking, fronthaul has shorter latency and wider bandwidth requirements when compared to backhaul. Today, the fronthaul transport solution is usually based on dark fiber – a point-to-point connection between the antenna site and the CRAN hub.

What changes with 5G?

With 5G, service providers start to introduce services using mid-band and high-band spectrum assets. While these bands provide massive capacity in 5G, they also introduce additional requirements on the transport network thus, requiring further investments. The decision for service providers then becomes choosing between expanding backhaul capacity or migrating to CRAN architecture with investments in fronthaul.

There are three key potential benefits with CRAN:

1. Antenna site simplification. Moving processing to a central office reduces the site footprint, power and battery backup requirements. This can be beneficial when:
   -  Expanding capacity on existing sites with limited space
   -  Acquiring new sites – especially important for micro/small cells deployment
2. Inter site coordination gains. When processing of multiple antenna sites is centralized in the same data center, it is easier to implement advanced inter site coordination functions such as carrier aggregation and uplink coordinated multi-point reception (UL CoMP).
3. Operational gains. CRAN can improve hardware (HW) utilization, since it can process radios from multiple sites and simplify capacity expansions.

However, to ensure these benefits, service providers need to look at their existing network and consider if it is fit to successfully migrate to CRAN. To illustrate the CRAN business case, let’s look at two different deployment scenarios.

Scenario 1 - suburban area with moderate traffic

In this first scenario, let’s consider the site sits in a suburban area with moderate traffic growth. The network is homogeneous, with only macro sites and no micro cells, and the service provider has microwave backhaul that can be upgraded in terms of capacity to support 5G.

Potential CRAN gains

• Antenna site simplification won’t be significant, as a cabinet will normally still be needed to store, for example, battery backup and backhaul equipment.
• Inter site coordination gains will also be relatively small, since the network is homogenous.
• Operational gains will be low, because the moderate traffic growth can usually be handled by adding additional frequency bands on existing sites.

CRAN costs

• The fronthaul investment in this scenario would be much higher when compared to the investment needed to expand microwave backhaul capacity.
• Since the service provider lacks access to data center locations, data center investments would be a high additional cost.

In this particular case, the cost outweighs the benefits of deploying CRAN, which means it is best to stay with DRAN and invest in microwave backhaul expansion.

Scenario 2 - dense urban area and high traffic

In the second scenario, let’s consider the site sits in a dense, urban area with high traffic growth. The network is heterogeneous, consisting of both macro and micro cells, and the service provider has access to existing fiber and a hub or data center facilities.

Potential CRAN gains

• As we have micro cells in this scenario, these antenna sites will have greater benefits from antenna site simplification.
• The heterogeneous network provides a good opportunity for coordination gains between micro cells and macro layers, resulting in higher RAN performance gains.
• Operational gains are also higher, since higher traffic growth requires more frequent site expansion and densification, which can be simplified with CRAN. There are also HW pooling gains across macro and micro/small cells.

CRAN costs

• The fronthaul investments in this scenario are much lower, since the service provider can reuse existing fiber investments.
• The same applies to data center investments, only to create additional capacity in an existing location.

As the picture shows, the gains clearly outweigh the costs in this scenario, so CRAN would be a sensible option. These are two select scenarios to illustrate what service providers need to consider when deploying CRAN. There are many other possibilities including combinations of de-centralized processing (L1, L2) for mmWave connected over higher layer split (HLS) to processing (L3) in CRAN hub.

Key considerations when opting for a CRAN deployment

1. Fronthaul considerations

In CRAN, fronthaul must deliver strict latency, capacity and synchronization requirements, which are significantly more stringent when compared to those of a DRAN backhaul connection. If the fronthaul network doesn’t fulfill these strict requirements (e.g., non-ideal fronthaul), this will impact spectral efficiency and more radios may be needed to compensate that. In the worst-case scenario, degradations can lead to loss of synchronization and instability which might jeopardize service completely.

There are different techniques to maximize fiber utilization in the RAN domain, such as radio CPRI cascading; multiplexing many radios onto a common, high capacity CPRI interface; and high band fronthaul sharing, where multiple radios are time multiplexed. Additional saving techniques available in the fronthaul toolkit when building a fronthaul network include:

Optical fronthaul

Dense Wavelength Division Multiplexing (DWDM) is a mature and well-established technology, capable of improving fiber efficiency by multiplexing multiple wavelengths over a single fiber. Utilizing DWDM in fronthaul networks reduces fiber cost and increases throughput.

Packet fronthaul

Packetized fronthaul can either be implemented in traditional radios, by converting CPRI interface radios into the enhanced CPRI interface (eCPRI) using fronthaul gateways, or natively in modern massive MIMO (M-MIMO) radios. Packet fronthaul is also an enabler for Cloud RAN executing on COTS servers with standard Ethernet interfaces. Packet fronthaul will substantially reduce the bandwidth and enable other Ethernet-based services over the same connection. For more information on this topic, read this blog.

Microwave-based fronthaul

Finally, there are also high-capacity microwave solutions designed to meet the strict requirements of fronthaul networks. This can be beneficial in CRAN as a complement to optical fiber. For example, if it is difficult to connect a rooftop site to the fiber fronthaul network, a microwave hop can be used to connect to another rooftop site and from there go to the data center via fiber.

At the end of the day, the optimal fronthaul solution depends on a number of factors such as fiber availability, CPRI vs. eCPRI radios and antenna site type, such as macro or micro cell.

2. RAN processing considerations

De-centralization of 5G core into edge or far-edge creates processing synergies with Centralized RAN. Service providers can choose between purpose-built HW or cloud infrastructure for RAN processing in the central office. Cloud RAN has a higher degree of elasticity for scaling up resources and can bring potential synergies from managing RAN workloads in the same way as core and edge compute workloads. Purpose-built HW on the other hand, will minimize energy consumption and footprint for RAN processing units, rectifiers and battery backup. More details on RAN processing options can be found in this previous Tech Unveiled blog.

Ericsson broad portfolio and expertise meets service providers where they are

There are different estimates of how many 4G/5G antenna sites are currently deployed with CRAN architecture, but the global average is single digit – meaning that over 90% of existing sites are deployed with DRAN architecture. However, CRAN will be a growing complement to DRAN in selected areas of the network.

Ericsson has the expertise and solutions needed to build both de-centralized and centralized sites. In CRAN, Ericsson offers both a purpose-built RAN and Cloud RAN portfolio; and when it comes to building fronthaul, we have a complete portfolio with WDM, packetized fronthaul and microwave solutions.

With Ericsson, customers get access to a broad portfolio that covers the complete network.