The business case for using residential broadband for mobile backhaul
An analysis by Bell Labs has found that using fixed access networks can significantly reduce the capital and operating costs for mobile backhaul (MBH). This is important at a time when mobile operators are facing continuous growth in bandwidth demand. A recent study has shown that, globally, demand will grow by a factor of 30 between 2014 and 2020. The same report also shows that densification of radio sites by adding smaller cells will be an important strategy to meet demand while maintaining operating margins. This densification of the mobile network results in higher capacity but, as a result, the backhaul network must also be densified.
With the mobile radio network evolving to heterogeneous networks (HETNET) composed of 2G/3G/LTE and soon 5G macro cells, small cells, and Wi-Fi hotspots, the cost effectiveness of the backhaul network becomes increasingly important. Fixed broadband access networks are ideally positioned to support mobile backhaul due to their dense presence in areas where wireless congestion and expansion will occur. As well as flexibility in terms of technology choice and future capacity upgrades, fixed access networks have a cost advantage over other MBH options.
In this article we will demonstrate how a shared fixed access/HETNET backhaul deployment lowers the total cost of ownership (TCO) for an operator offering both fixed and mobile services.
Mobile backhaul options: technologies & architectures
Many fixed operators are implementing or planning major upgrades of their access network. To satisfy increasing peak capacity demands, fiber is being brought closer to the customer premises and ultimately into the premises. Fiber-to-the-x (FTTx), with “x” indicating a reference point located between the central office (CO) and the customer premises, uses various access technologies (see figure 1). We focus on technologies with speeds of 150 Mb/s or higher to backhaul radio cells deployed in urban areas.
Broadband solutions based on fiber-fed distribution point units (DPUs) that are positioned deep in the network at fiber-to-the-distribution point (FTTdp) or fiber-to-the-building (FTTB) locations, leverage G.fast technology to reach 1 Gb/s (aggregated downstream + upstream) from the DPU to the subscriber over existing copper drops. FTTx predominantly uses Passive Optical Network (PON) technologies, such as GPON or XGS-PON, from the CO to the DPU. Fiber-to-the-home, FTTH, also generally uses PON. PON’s passive splitting architecture typically connects up to 32 or 64 endpoints (premises or DPUs) that can be serviced with a single feeder fiber, with all end points sharing the PON capacity.
In areas where a deep fiber FTTx network exists, fiber-based backhaul of the radio cell is an obvious choice: the dense geographical presence of available drop cable or ducts and the inherent overprovisioning of passive fiber assets (fibers and splitters in distribution and feeder areas) can be leveraged for connecting fiber to every radio cell (FTTCell) deployed in the serviced area.
Figure 1: Technologies and architectures for mobile backhaul solutions
In more traditional solutions, the radio cell is backhauled via microwave (MW) to a nearby macro cell location (MC-RAN) or other aggregation location equipped with dedicated fiber for the transport of MBH signals. It can either be a point-to-multipoint (P2MP) MW solution (sub 6 GHz frequency domain, near-line-of-sight (n-LOS)) or a point-to-point (P2P) MW solution, requiring line-of-sight (LOS) in the 60 GHz frequency domain. These solutions are generally quite expensive with limited capacity, dependent on weather conditions (e.g. typically 150 Mb/s for P2MP MW). The advantage of MW backhaul is that it can be implemented quite quickly.
Cost comparison of different backhaul solutions
We compared the following solutions for connecting metro cells (at comparable data rates):
- MDU FTTB. Similar to the way existing twisted pairs are leveraged inside multi-dwelling units (MDUs), an antenna positioned on the façade or the roof of a building can be connected to a DPU located in the basement and G.fast used over the in-building copper infrastructure.
- GPON FTTCell. Fully converged backhaul/access solutions leveraging GPON, where residential subscribers (FTTH) and metro cells (FTTCell) share the PON capacity and QoS mechanisms ensure that MBH delays and bandwidth requirements are met.
- P2P FTTCell. P2P Ethernet fiber solution dedicated to the metro cell (1 Gb/s) but leveraging the OSP (Outside Plant) passive assets with common CO OLT equipment.
- P2P fiber MC-RAN. The metro cell, serving a hot spot, is backhauled with optical Gigabit Ethernet to a nearby macro antenna in the same region. In this case two fiber drop installations are required: one to connect the cell site to the nearest fiber connection point (FCP) and the other to connect the macro cell to its nearest FCP. End-to-end backhaul connectivity is realized via spare distribution fiber(s) in the OSP. In contrast to P2P FTTCell, there is no convergence in the CO.
- P2MP MW: n-LOS microwave in the sub-6 GHz frequency domain.
- P2P MW: LOS microwave in the 60 GHz frequency domain.
While it is relatively straightforward to assess the CAPEX for dedicated backhaul solutions (the sum of equipment and installation costs), FTTx-based solutions share infrastructure deployed in the first place for providing residential services. We assess the backhaul cost for a metro cell leveraging OSP assets by regarding a cell site as another connected premise for which we can calculate the deployment cost per connected premise. This approach is justified as metro cells will represent less than 1% of connected broadband subscribers. In a typical urban setting (2000 premises/km2) and where a 50% service take rate is targeted, for a cell radius of 200m, we take into account one metro cell per 128 connected subscribers, or one additional metro cell per 2 PONs. With these assumptions and with CAPEX being the dominant contributor to the MBH TCO, Figure 2 gives the relative CAPEX costs for the different backhaul solutions.
Figure 2: CAPEX contributors for various backhaul solutions, normalized.
It shows that backhaul solutions based on a nearby FTTx network (terminated at the CO) result in the lowest CAPEX cost. The MDU FTTB solution (with backhauling via in-building copper) comes out as the lowest cost option, whereas a solution with P2P fiber to the CO is 30% more expensive than connecting a GPON. The other backhaul options are 2 to 3 times more expensive. Note that P2P FTTCell in our scenario still requires 25% more feeder fiber overprovisioning compared to a GPON solution.
Leveraging spare/dark distribution fibers to connect small cell sites to macro cell sites with P2P Gigabit Ethernet comes at a lower cost than microwave backhaul. As long as the last drop stays below 75m, MW backhaul will be more costly, driven by the relatively high cost for the microwave equipment. Note that we assumed the MW backhaul to operate in a license-free spectral region.
Converged capacity upgrade of FTTCell and FTTH
Where a converged access/backhaul solution serves both fixed subscribers and radio cells, GPON capacity will eventually be exhausted. Traffic forecast modeling (1) resulting in +20% compound annual growth rates shows that GPON can cope until 2020 for PONs serving 100 active users. But GPON capacity may also require a drastic upgrade if the operator decides to upgrade the capacity of the radio cell or switches to new RAN technologies (e.g. 5G).
Figure 3: initial+upgrade CAPEX for various backhaul solutions, normalized
In this section we demonstrate the unparalleled scalability of access in case of upgrades. PON capacity can be enhanced to symmetrical 10 Gb/s (XGS-PON) or 40 Gb/s (TWDM-PON). Different options exist to upgrade a converged broadband access-MBH solution. Operators have flexibility in choosing how capacity upgrades can be introduced and customized for each service area.
- A first option is upgrading GPON-based FTTCell to P2P connections (FTTCell to P2P, providing 1 Gb/s symmetrical backhaul capacity) or to XGS-PON dedicated for the radio cells (FTTCell dedicated XGSPON, e.g. providing 1.25 Gb/s to each cell by connecting 8 radio cells to a single PON). Both cases require touching the OSP and consume extra feeder/distribution fibers.
- A more holistic approach that involves residential and business subscribers, is a converged XGS-PON upgrade. It leaves the OSP untouched by activating an XGS-PON OLT port in the CO and by wavelength multiplexing the GPON and XGS-PON signals on the same PON feeder fiber. The radio cells, as well as a selection of premium/enterprise users are then serviced by replacing GPON ONTs with XGS-PON ONTs.
The latter upgrade option provides already an attractive TCO in the case where 10% of the premises (FTTH+FTTCell) are upgraded to XGS-PON and share the cost. It nearly equals the cost of upgrading FTTCell to a P2P solution (see figure 3), with the additional benefit of having upgraded selected FTTH users. And it is still more cost-efficient compared to dedicating XGS-PON to the radio cells only or any MW-based solution upgrade based on software upgrades, installing new MW links or switching to higher capacity hardware.
Above all, fixed networks allow to allocate a capacity share beyond 1 Gb/s to MBH with a better cost per Mb/s. By opting for a GPON/XGS-PON solution, a fully symmetrical 10Gb/s converged access/backhaul solution is achieved, with the operator fully in control of when and where to spend the associated upgrade CAPEX, to the benefit of both its fixed and mobile subscribers.