Alcatel-Lucent

Tailoring baseband processing and backhauling for their requirements and existing assets helps wireless service providers meet rapidly changing requirements and reduce total cost of ownership.

Highlights

• lightRadio provides the flexibility to deal with change and growth
• Baseband deployment and backhaul options can be matched to fiber availability
• Centralized baseband processing has key advantages over other approaches

Versatility and reuse of assets
With their ability to support different baseband processing and backhaul configurations, Alcatel-Lucent lightRadio products help wireless service providers deal with change and growth. lightRadio products:

• Can be used with IP, fiber and microwave backhaul — whether assets are leased or owned. They also enable Digital Subscriber Line (DSL) and Fiber-to-the-Node (FTTN) copper backhaul.
• Can be used with varying levels of dark fiber availability in the first and second miles.
• Offer remotely programmable baseband processing that supports different Global System for Mobile Communications (GSM), Wideband Code Division Multiple Access (W-CDMA) and Long Term Evolution (LTE) traffic mixes on the same digital hardware.
• Provide multiband capabilities that reduce the number of pieces of equipment on a tower without increasing the risk of failure.
• Include both multi-sector macro and single-sector metro (pico) offerings with identical features, seamless interoperability, optimized hand-off and interference avoidance.

The lightRadio product family also supports intelligence and processing in multiple locations. Any given network can have a mix of architectures. And wireless service providers can change between architectures without writing off assets.

lightRadio baseband deployment architectures
Figure 1 illustrates the lightRadio baseband deployment architectures.

Figure 1: lightRadio supports multiple baseband deployment architectures

As described in Table 1, all-in-one and conventional baseband processing offer different benefits.

Table 1: All-in-one baseband processing differs from conventional baseband processing

With a centralized baseband unit (BBU), digital baseband processing is performed at a distance from the remote radio heads (RRHs) and antennas. There are three approaches to centralized baseband processing:

• Clustering puts a stack of conventional BBUs in a central location. Each BBU is connected to an RRH using the Common Public Radio Interface (CPRI) over fiber. We refer to this as CPRI interconnect.
• Pooling treats digital resources as a single, grouped resource. This allows lightly and heavily loaded base stations to be load-balanced across the pool.
• Cooperative shares information among different base stations to improve capacity and performance. This approach takes advantage of coordinated multipoint transmission (CoMP). CoMP brings wireless service providers a set of cooperative features that become more effective, easier to implement and less costly with centralized baseband processing.

IP backhaul network options
Where IP backhaul (post-baseband processing) is used, delivery over an IP mobile backhaul network is common. Figure 2 illustrates lightRadio IP backhaul network options.

Figure 2: lightRadio supports different IP backhaul network options

From a transport perspective, IP backhaul is relatively easy. IP traffic has relatively low bandwidth requirements and is not sensitive to latency or jitter. In addition, IP backhaul traffic can be carried over almost all broadband networking media, including microwave links, multiple passive optical network (PON) variants, DSL, hybrid fiber coax (HFC) networks and fiber optics.

However, transport of I/Q sample signals between an RRH and a centralized BBU is not so simple. The bandwidth requirements before compression are roughly 20 times the IP bandwidth. And the traffic is extremely sensitive to latency and jitter. In most cases, broadband networking media other than fiber are either technically or economically infeasible.

Centralized baseband processing and backhaul options
Where baseband processing is centralized, the CPRI interconnection from the RRHs is assumed to be on a separate fiber pair — or color of light (λ).

The following cases describe lightRadio backhauling and CPRI interconnect for scenarios with progressively less fiber. Table 2 summarizes the cases using the following definitions:

• First mile: The span between a remote base station and the first building housing aggregation equipment such as switches and routers — typically a central office (CO) building.
• Second mile: The span between that CO or point of aggregation and the next higher level of aggregation — a metro point of presence, for example.

Table 2: Dark fiber availability influences baseband processing and backhauling approaches

Case 1: Abundant dark fiber in the first and second miles
With abundant fiber, service providers can take advantage of centralized baseband processing. Centralized baseband processing typically brings traffic from dozens of base stations back to a metro site or point of presence. There are three main benefits to this approach:

• No equipment is required at the base of a radio site. This reduces space and cost requirements.
• Centralized equipment simplifies maintenance and upgrades for digital processing equipment.
• New techniques, such as joint processing coherent CoMP, can be used to improve capacity and performance.

Other factors also need to be considered:

• Scalability: Locating central processing at the first point of aggregation creates scale issues. If centralization gains are expected to come from averaging the traffic from heavily and lightly loaded base stations, the central site should aggregate traffic from base stations serving different demographics and with different time-of-day peak loads.
• Cluster size: Potential gains from CoMP increase when all of the “interfering” base stations are within the same central processing cluster. Larger clusters can typically deliver larger gains. A central processing cluster should aggregate at least 15 base stations to achieve most of the potential CoMP gains. Clusters of 30 or more base stations are preferred.

Figure 3: Baseband processing and backhauling in a rich fiber architecture

Case 2: Abundant dark fiber in the first mile, little in the second
Often, wireless service providers have abundant dark fiber in the first mile but much less in the second mile. With the high cost of adding large numbers of optical interfaces at the RRH, it is generally better in these cases to act as if fibers were scarce in both the first and second miles. If separate fibers from the RRH are required, then Wavelength Division Multiplexing (WDM) can be used at the remote site.

Figure 4: Baseband processing and backhauling in a site-based WDM architecture

If separate fiber backhaul is required right through the first mile — for example, to increase robustness or reduce single points of failure — then WDM at the CO level is recommended.

Figure 5: Baseband processing and backhauling in a CO-based WDM architecture

The choice of coarse or dense WDM depends on the number of signals to be carried and the number of available fibers in the second mile:

• If sufficient fiber is available, then multiple fibers carrying Coarse Wavelength Division Multiplexing (CWDM), which can typically support eight colors, will lower costs.
• If fiber is insufficient for CWDM, the more expensive Dense Wave Division Multiplexing (DWDM) option may be necessary.

Case 3: Scarce dark fiber in both the first and second miles
Most wireless service providers have scarce fiber in both the first and second miles. To help them make the most of the fiber they have, lightRadio products support both aggregation and compression of CPRI data when centralized baseband processing is used. With this approach, CPRI data can be carried over a single 10 Gb/s fiber pair. And the full loads from all base station sizes anticipated in the next five years can be carried over two fiber pairs.

As in Case 1, central processing clusters should be as large as possible. However, in some cases, the topology of available fiber may make “hub COs” a better choice for centralized processing than higher level metro aggregation points.

Figure 6 illustrates the three scenarios in which wireless service providers can deploy centralized baseband processing over top of existing assets:

• Using the spare fibers in an optical ring to create a new optical ring for CPRI interconnect (3A)
• Putting CPRI signals onto a color of light that shares a fiber with EPON, GPON, 10GPON in the second mile (3B)
• Using the spare fibers in an FTTN or HFC network to create a CPRI overlay (3C)

Figure 6: Baseband processing and backhauling in an architecture with scarce fiber

All three scenarios take advantage of a new Alcatel-Lucent device called a CPRI MUX. Located at the remote site, a CPRI MUX aggregates traffic from many different radios on the site and encapsulates it for transport over the minimum number of optical interfaces. This allows the vast majority of sites to be contained within a single 10 Gb/s optical interface.

Case 4: No owned dark fiber
When considering a macro cell with no owned dark fiber:

• “Owned” means available to the wireless service provider at marginal cost.
• “Dark” means it is a spare fiber pair, available to be lit up and not part of a switched or routed network.

In this case, aggregate IP traffic will not exceed 1 Gb/s. That means options, such as leased fiber backhaul and owned packet microwave backhaul, can be used.

Leased fiber backhaul
Fiber can be leased on a per-Mb/s (common), per-λ (rare) or per-fiber (dark fiber) basis.

In this case, either the all-in-one or conventional distributed baseband processing options are appropriate because they minimize backhaul leasing costs. lightRadio reuses the same digital building blocks and software in different architectural configurations so either option, or a mix, can be used.

Figure 7: Baseband processing and backhauling in a leased fiber architecture

Owned packet microwave backhaul
Owned packet microwave backhaul can be single-hop, multi-hop or more complex arrangements with rings and spurs.

Figure 8: Baseband processing and backhauling in a microwave backhaul architecture

Owned fiber with Ethernet aggregation
This case is different from other owned fiber cases because the infrastructure is typically organized so that many fiber pairs home to carrier Ethernet switches and routers. In these networks, processed IP traffic can be transported but CPRI-formatted antenna signals cannot. This is based on the ability guarantee the bandwidth, latency and jitter levels to which traffic will be exposed. Again, either the conventional distributed baseband processing or all-in-one option is appropriate.

Figure 9: Baseband processing and backhauling in an Ethernet aggregation architecture

When base stations are near a PON infrastructure, it can provide effective backhaul for remote, site-based broadband options. However, PON is not suitable for macro cell CPRI interconnect^[1].  Centralized baseband processing simply has too much constant-bit-rate traffic. A large macro base station could generate 30.7 Gb/s of uncompressed CPRI traffic, while typical PON systems reach their maximum at 10 Gb/s per PON segment.

Base stations served by copper pairs
In this case, the wireless service provider has a base station served by copper pairs, such as E1 or T1. However, fiber is also deployed in the access network to within a kilometer or less from the base station.

Figure 10: Baseband processing and backhauling in an architecture with copper pairs

The current copper pairs can be converted to channel-bonded very-high-bit-rate DSL (VDSL) with vectoring and Phantom mode. This approach delivers quite reasonable performance for small and medium-sized base stations with two pairs or more per RRH.

However, this option is not compatible with centralized baseband processing due to limited available bandwidth and asymmetrical traffic patterns. Latency is also a concern as DSL’s approximately 3 ms of latency is borderline for viable baseband processing performance.

The flexibility to adapt
Backhaul makes up a large proportion of most wireless service providers’ operating and capital costs. Higher bandwidths over fiber can enable centralized baseband processing. However, most sites are not currently served by fiber — a situation that is not expected to change any time soon. As a result, new solutions need to support a variety of base station deployments.

lightRadio is designed to support a variety of baseband processing and backhaul options. For example, wireless service providers can take advantage of CPRI interconnect and backhaul. They can also take advantage of centralized baseband processing beyond the obvious scenario — where they already own dark fiber between the base station and a centralized processing site.

To contact the authors or request additional information, please send an e-mail to techzine.editor@alcatel-lucent.com.

Footnotes

1. ^[1] PON fiber can be used for CPRI interconnect assuming that the CPRI traffic uses dedicated wavelengths on the PON fiber that is not interfering with the native PON wavelengths. ↩