To license or not to license: Building reliable, high-bandwidth wireless networks

There is a lot of debate in the wireless telecom industry about the role of unlicensed spectrum vs. the role of licensed spectrum when building wireless networks. Licensed operation provides the security of interference-free operation while unlicensed technologies have facilitated commoditized equipment availabilities that have essentially 'enabled" a number of market segments through sheer cost-effectiveness of end user equipment, not to mention the impressive technical performance attributes available.

This article examines a general network architecture and positions technology solutions within it in an effort to illustrate the optimal application of each.

General metro network architecture

A typical metropolitan network or private multi-building LAN network can be crudely split into three segments as shown in Figure 1, namely:

• Core network layer (applicable to public networks, not so much to private multi-building LANs; usually fiber based in the case of high-speed data networking)

• Aggregation or backhaul network layer

• End user or end customer access layer

Figure 1: Typical network layering showing criticality/availability and data rate dependencies

Wireless technologies usually find a home in the lower two layers of this network structure model since the core layer usually employs data rates and availability performances that are beyond that which can be delivered with wireless technology (typically the network cores are 99.999% available or higher and fully hardened).

As shown in Figure 1, moving from the outer end-user (or end customer access) access layer toward the metro core generally is accompanied by increasing aggregation of access traffic onto progressively larger and larger bandwidth connections. As the bandwidths go up, and the number of end users aggregated increases, the more painful link outages become.

There are several different general types of networks that can be similarly viewed. Table 1 shows this view of the network. This table also illustrates the differences between mobile/portable data access networks (i.e., Wi-Fi hot spots), conventional mobile voice networks and fixed wireless access (FWA) networks used for business access services.

Table 1: Performance and cost demands for various network layers

The table underscores conventional wisdom in that availability (dependability and reliability) demands go up as bandwidth demands increase. Additionally, when a wireless link is fixed, outages or modifications to a wireless link to solve interference (or other) problems is time consuming, costly and affects churn. In contrast, access layers in which mobility is a performance feature can "transfer" the link performance onto the end user, who "repairs" a poorly performing link by physically moving to a site with better reception qualities. Additionally, poor access link performance affects a low number of end users, in contrast to aggregation/backhaul or core link performances, which can affect hundreds or thousands of end users. For this reason, any type of access network relies on high bandwidth, highly reliable aggregation/backhaul and core infrastructure.

Choices of wireless technologies: What goes where in the network

There are a number of basic choices in operating frequencies for wireless networking. These can be crudely broken down into various licensed and unlicensed technologies. As shown in Figure 2, unlicensed technologies at lower frequencies (i.e., 2.4 and 5.8 GHz) cannot provide reasonable availability performance because the threat of interference essentially renders these technologies unreliable and unpredictable. In contrast, at the other extreme, licensed technologies provide reliability and predictability through interference-free operation. All wireless technologies have to confront the challenges of line-of-site blockages and propagation factors.

Figure 2: Relative reliability and predictability of various wireless implementation technologies

Wireless links that are installed in clear, unobstructed LOS conditions can offer highly predictable and reliable performance akin that available from fiber optics. Generally, these are designed based on long-term statistical methodologies that have been thoroughly time-proven. When blocked-LOS or non-LOS conditions are attempted, it is not possible to achieve high reliability or predictability since propagation through objects (trees, buildings, etc) is highly unpredictable, and so are the wireless links implemented with these features. Additionally, initial install success on non-LOS links also presents a serious exposure due to the inability to predict the overall propagation performance.

Interference is a nasty beast that can strike anywhere. Although spectrum congestion certainly affects the odds of this happening, it is a very real threat regardless of where the deployment is undertaken. This is a little like a car accident--it can happen at anytime.

Mapping the various frequency-driven technology solutions onto the network requirements of Table 1 can be done by comparing the solutions in terms of bandwidth and cost (see Figure 3). Essentially, this figure illustrates that 60 GHz and 24 GHz unlicensed along with licensed PTP links are a good fit for the aggregation/backhaul and core network layers. Since the core of these networks generally has bandwidth demands that demand fiber optic data rates, wireless technology does not find widespread application (with the minor exception of high-speed free air optics and some 60 GHz systems that are able to operate at 622 Mb/s and higher).

Additionally, the lower cost of unlicensed 2.4 and 5.8 GHz technologies make them well suited for very short-range access applications. The low cost of mobile handsets has similar obvious application to the access layer.

Figure 3: Bandwidth comparisons of various frequency/technology options

Some wireless licensing basics

So, comparisons aside, what does it cost to achieve the interference-free security and dependability of licensed operation? Licensing generally falls into two categories, namely:

• Area license: This is required when a multipoint deployment is undertaken and many sites are connected. In some cases, area licenses also allow mobile or portable applications. These are purchased by the operator through auctions or through applications. Usually their value is a function of households passed (or similar) coverage-oriented value scale.

• Site license: This is required where a point-to-point link is installed between two fixed sites. Costs are generally $100 - $200 per year.

Summary

Licensed and unlicensed technologies are well suited for different applications within the various layers of a typical network as summarized in Table 2.

Table 2: Summary of the applicability of various technologies and frequency choices to the layered network model

Erik Boch is co-founder and Chief Technical Officer of DragonWave Inc.

Visit DragonWave Inc. online.