As mobile networks explode with data traffic, operators can tap different backhaul architecture, self-organizing networks, and inexpensive spectrum to clear up the small-cell backhaul bottleneck.
Smart phones and tablets are becoming devices of choice for consumers and road warriors. Whether they are at the local mall, Starbucks, or Panera, consumers have come to expect that when they use their portable device that they will be able to use social media, surf the net, and utilize applications that were specifically designed to entertain them and make their lives easier.
Today, we automatically assume that we can access applications and the Internet if we have a connection. But with the rapid adoption of smart phones and tablets, will the network be able to handle it? Or will our applications just time out, stall, or take forever to run? Operators are still trying to figure out how to meet the anticipated demand.
One solution to the evolving wireless broadband capacity crunch is a heterogeneous access network comprising macrocells, plus a range of small-cell solutions: microcells and picocells, carrier Wi-Fi, and femtocells to add capacity and performance. Wireless operators are starting to implement small-cell access solutions, though adoption has slowed down due to the lack of cost-effective backhaul solutions. Fiber has proven to be too expensive, and microwave isn’t technically feasible in many urban environments where line-of-sight is not possible.
There are three major design elements that need to be addressed to solve the small-cell backhaul challenge: architecture, self-organizing networks, and spectrum.
Currently, mobile backhaul solutions such as microwave and fiber utilize a point-to-point (PTP) architecture where basestations have full utilization of dedicated backhaul bandwidth. PTP is provisioned based on peak volume, so during off-peak times the excess spectrum is idle.
Point-to-Multipoint (PMP) is the ability to backhaul several basestations through a single backhaul hub by creating multiple links that share the hub bandwidth (Fig. 1). PMP backhaul is intelligent, based on sharing resources, and ideal technically and economically for small-cell backhaul.
Technically, traffic patterns on for small-cell basestations are very different from macrocell basestations. Compact, below-roofline basestations such as microcells or picocells transmit at low power and are mounted at low height. This means their coverage area is small. Small coverage area means less interference with adjacent cells, and that’s why compact basestations provide up to double the capacity of macro basestations.
Macrocells have a low peak-to-average traffic ratio whereas small cells have a high peak-to-average traffic ratio. On average, there are fewer users served by small cells, but the performance is better (i.e., better download and upload speeds). When a small cell peaks, it does so for a very short period of time but can sustain rates above those of a macro cell.
Furthermore, traffic patterns for data services are not as predictable as those of voice networks. For example, voice networks have busy hours that typically happen at known times during the day where subscribers drive up the traffic on several basestations at the same time. This is different in data networks where traffic patterns are not predictable. Data consumption can happen at any time and is different between basestations.
Based on the described traffic patterns, it would be extremely inefficient to dedicate bandwidth to each compact basestation independently, especially if the dedicated bandwidth is designed to match the peak traffic performance that can only be achieved for very a brief moment with small-cell architectures. Given that peak traffic on any basestation can only occur when network utilization is low—that is, when adjacent basestations aren’t very busy—it becomes advantageous to use PMP backhaul.
In this case, several basestations share the bandwidth, which is fine given that they don’t peak at the same time, and if they do, it’s only for a very brief period of time where traffic on adjacent basestations is low. PMP configuration is made more efficient by using “dynamic bandwidth allocation,” which shifts resources from a basestation with low traffic requirements to one with high traffic demand. This enables optimization of bandwidth to serve the basestation with highest traffic requirements at any particular moment in time.
Self-Organizing Radio Networks
Today’s cellular networks are evolving to be fully automated. Built-in intelligence provides the ability to self-configure, self-optimize, and self-mange to provide high network quality and a satisfying subscriber experience. These new self-organizing networks (SONs) help bring down the costs of network operations by reducing time-consuming manual work by shifting the hands-on configuration, management, and on-site maintenance to a fully automated process:
- Self-configure: In mobile backhaul self-organizing networks (B-SONs), backhaul system components are plug-and-play and automatically connect and configure each remote backhaul module to the backhaul hub.
- Self-optimize: B-SONs are designed to continuously adapt to changing environments. Radio-frequency monitoring is systematic, allowing the backhaul systems to automatically reconfigure to adapt to changes in the network such as the addition of new backhaul nodes or thicker vegetation in summer months. Dynamic resource allocation distributes bandwidth to minimize traffic bottlenecks, and intelligent traffic control guarantees quality of service for mission-critical traffic to ensure high performance.
- Self-manage: B-SONs automatically detect service degradations or a single failure anywhere within the system and reconfigure themselves around the point of failure.
SONs are especially well suited for urban areas where there will be dense small cell deployments. In fact, SONs make it technically and economically feasible for operators to quickly increase capacity in emerging hotspots. Benefits include faster time-to-market for new services, faster return on investment, and more efficient utilization of existing network assets, increased operational efficiency, and improved end-user experience.
Spectrum is very expensive. New mobile backhaul solutions are utilizing sub-6-GHz time-division duplex (TDD) with spectrum management to drive costs down. TDD spectrum in 2.3-, 2.5-, and 3.5-GHz bands is widely available in many countries at far lower prices than access spectrum.
In recent auctions in Germany, beachfront 800-MHz digital dividend spectrum sold at over 74 euro-cents per megahertz per head of population (MHz-PoP) whereas 2.5-GHz unpaired (TDD) spectrum sold for about 2 euro-cents per MHz-PoP. Similar and even lower prices were fetched for TDD spectrum in other European and Asian markets.
This is recognition of the fact that an access network in high spectral bands requires greater capital expenditure as cells shrink and it becomes particularly challenging to provide indoor coverage. In addition, licensing in the sub-6-GHz TDD band is different from that in typical microwave bands where licensing is on a per-link basis according to channel bandwidth.
Most countries license sub-6-GHz spectrum after the license cost is paid for a long period of time, typically 20 years. This suits well the model of deploying backhaul for a large number of small-cell basestations where the cost of spectrum can be amortized over a large number of small cells.
Because RF performance is key to providing high spectral efficiency, active spectrum management in the backhaul network becomes a critical differentiating factor that allows the operator to squeeze maximum capacity out of limited spectrum holdings. Active spectrum management includes a continuous and automatic process that characterizes the radio-frequency performance of every link in the network (Fig. 2) and determines inter-link interference between in a network of many PMP backhaul clusters. (A cluster is a backhaul hub servicing several remote backhaul modules.)
This allows the network to “self-configure” by coordinating the operation of the backhaul nodes in a manner whereby high inter-link interference can be avoided. Lower interference leads to higher signal quality, which in return allows for higher capacity and spectral efficiency.
Active spectrum management also acts to reduce the operator’s operational expenditure in a number of important ways. For one, it takes the labor and guesswork out of the site planning process by eliminating the need to hunt down sources of interference and manually coordinate the frequency assignments of backhaul links. In fact, some SON features enable active spectrum management: inter-link interference characterization.
Efficient spectrum management enables higher spectrum utilization, increasing capacity and reducing backhaul spectrum requirements. It provides higher signal fidelity by allowing the backhaul systems to adaptively lock on a high modulation and coding scheme. More importantly, it allows for scalability of the backhaul network to enable deployments of large numbers of small cells to address the operator’s capacity needs and meet explosive data traffic growth.
The combination of point-to-multipoint architecture, built-in intelligence in SONs, and spectral efficiency creates a compelling technical and business case for small-cell backhaul. These technologies are critical to reduce the cost of capacity ($/bits/s), which is fundamental for network operators to maintain and even expand their profitability. Small cells do provide much higher capacity gain (2x bits/s) than macrocells, in addition to other performance benefits, but their widespread adoption has been limited due to challenges in providing cost-effective backhaul (Figure 3 and Figure 4).
Now, with these technologies, it is possible to roll out a network of small-cell basestations while reducing backhaul costs by more than 50% over current techniques such as fiber. Heterogeneous networks then become a possible reality, not merely an interesting technical concept where everyone agrees on its superiority but no one can implement in a profitable way.