There is a near universal consensus among mobile operators that small cells are the way of the future – the only solution that will enable them to provide the capacity they need to meet the expectations for the fast growing numbers of smartphone users. Small cells allow operators to use their existing and new spectrum assets more intensely by increasing the capacity density (the gigabits per second they can squeeze out of a square mile).
But this is where the agreement ends. It is not yet clear how small cell deployment will look like, and there may be substantial differences across regions and operators. How dense will small cell deployments be? Will small cells have 3G and Wi-Fi in addition to LTE? Or will 3G small cells come first? How will interference with the macro cells be managed?
Backhaul is one of the areas where there is still little clarity. No single backhaul solution will suffice. Fiber is not available or cost effective in most locations. Wireless backhaul will be used in most cases, but operators will have to pick the best solution for each small cell. And this is where the uncertainty lies. Is there a way to limit the number of solutions needed – and if so what is a simple, reliable decision process to pick the right solution for each small cell?
I have presented the research I did on this topic in a recent report, both from a performance and financial perspective: operators face multiple tradeoffs in selecting small cells backhaul to minimize costs while meeting their performance requirements.
They cannot have it all. Fiber or other wireline alternatives are ideal where available and not too expensive (in our model a lease of $2,500 per year makes fiber the preferred option, assuming moderate trenching costs). With wireless backhaul, there are two solutions that we expect to prevail:
–Sub-6 GHz non-line-of-sight links, mostly using a point-to-multipoint architecture and licensed spectrum.
–Millimeter-wave (60 GHz and e-band) solutions that require line of sight and use a point-to-point architecture.
The two solutions are almost diametrically opposite in terms of benefits and challenges. Sub-6 GHz NLOS backhaul is capacity limited, as spectrum is very tight and expensive, but allows operators to reach small cells that do not have an unobstructed path to a macro cell or another small cell. Millimeter-wave backhaul provides much more capacity, but only works where there is LOS.
From a performance point of view, operators may decide to deploy LOS millimeter-wave backhaul wherever they need wireless backhaul and LOS is available, and use sub-6 GHz NLOS for those cells that are not within LOS or that cannot be reached by a relay link to establish LOS.
Some operators are following this path, but others have a preference for sub-6 GHz NLOS links. NLOS backhaul provides less capacity, but it is more flexible and less expensive to deploy and maintain. So they plan to use it wherever possible, and switch to millimeter-wave where they hit the capacity limitations of sub-6 GHz.
What is the best approach? We looked at this tradeoff in two ways: the total cost of ownership per link (Figure 1) and per Gbps (Figure 2). If you take the TCO per link, sub-6 GHz PMP solutions are less expensive, as less equipment is needed (if you have five small cells backhauling to a macro cell, you need six terminals for PMP, and 10 for PTP), the installation is easier, and leasing costs are lower. If you look at the TCO per mbps, however, millimeter-wave solutions provide the best value, as each link has higher capacity.
The choice between the two solutions then depends not only on the availability of LOS, but also on the density and capacity requirements of the small-cell deployment (Figure 3).
For a network with densely packed low-capacity small cells, millimeter-wave solutions are preferable because they allow higher spectrum reuse and provide the capacity needed for different topologies (star, ring, hub-and-spoke, or hybrid).
A low density of 3G low-capacity small cell with more limited capacity requirements is likely to be served more cost-effectively by NLOS backhaul that provides more flexibility in reaching small cells. Low density of cells makes the more limited spectrum reuse of sub-6 GHz spectrum an acceptable tradeoff if capacity demands are not high.
A low density, low capacity small cell deployment may benefit from a NLOS solution because it allows the operator to increase capacity easily within the same area simply by adding new small cells that share the same backhaul hub and because as cells are further apart from each other the availability of LOS tends to be lower. If there is LOS, however, millimeter wave backhaul provides more capacity at a similar cost, because in a low-density network the cost advantages of PMP are substantially lower as they depend on multiple small cells sharing the aggregation hub.
Finally, in a high-density, high-capacity network, LOS is well suited because short links are more likely to have LOS and it offers higher spectrum reuse. However, NLOS can provide a relatively higher spectrum reuse in such environment (if each PMP aggregation hub serves a region within a small radius) and hence higher capacity, while keeping down per-link costs, thus providing an attractive solution for operators that are more aggressively trying to keep costs down.
The analysis shows that after taking into account spectrum and LOS availability, picking the cheapest solution – i.e., using the per-link TCO – is not always the best choice. Rather, performance and capacity requirements need to be jointly taken into account along with the TCO to optimize capex and opex for the specific requirements of each small cell.