Author: Alan Solheim
VP Product Management, Dragonwave, Inc.

May 23rd, 2007

Most of the discussion about WiMAX technology and economics has been centered around the radio access network (RAN) and customer premise equipment (CPE)/handsets. For services to be deployed profitably, however, the economics of the entire network, including backhaul infrastructure, and installation and leasing costs, must be considered. In this larger view, it quickly becomes clear that the RAN is only a portion of the overall picture and costs.

The total deployed base station cost is made up of equipment and installation costs and the net present value (NPV) of the ongoing sector antenna lease and maintenance costs. As the cost of base station hardware is reduced by technology development and increasing production volumes, the deployed cost becomes more and more dependant on the latter two costs. A view of the relative contribution of these costs over time is shown in the figure below using a five-year NPV calculation of recurring costs.

Backhaul costs are dependant upon the technology chosen as well as network architecture. The backhaul bandwidth that must be provisioned per WiMAX base station will grow from a few Mbps initially to up to more than 100 Mbps when new services are widely adopted. This bandwidth requirement will be further increased by 20% to 100% by the type of network redundancy employed (ring/mesh or 1+1). In order to provide a balance between scalability and reliability, multi-tier network architectures with access links connected to aggregation hubs are typically employed in such network architectures. The technology choice in the first tier of backhaul is driven by the average bandwidth per base station. The technology choice in the aggregation layer is driven by network topology and number of end sites per aggregation hubs.

Initially, it may be possible to use in-band backhaul when bandwidth per base station is low. The requirement to deliver services to the end user will quickly drive up the bandwidth needed per base station and mean that any spectrum allocated to an in-band backhaul solution will need to be recovered for the RAN. Another way to say this is that WiMAX spectrum is too valuable to use for backhaul and must be reserved for providing service to the end user and driving revenue onto the network.

T1/E1 leased circuits are a simple option but are not - able to scale to the required bandwidth at a cost point that is competitive. Lease costs in North America are $200 to $400 per T1 per month and $1500 to $3000 per DS3 per month. When this is added to the equipment cost required to convert the IP interfaces of WiMAX base stations to the TDM format it becomes clear that, at any base station capacity above a few Mbps, - this is not a viable option.

Providers who choose these alternatives initially will have to pay the price of churning their networks to install the alternatives required to deliver increased bandwidth. Given that the installation/reconfiguration cost is a significant portion of any deployment, as is seen from the base station total costs above, it is very often more economical to deploy a scalable backhaul solution even if the day one price is slightly higher.

Other solutions for backhaul that can deliver the required bandwidth are point-to-point microwave and fiber-based Ethernet. The advantage of fiber-based Ethernet is that the bandwidth is effectively unlimited. The bandwidth that can be sourced by a single base station is strictly limited by the amount of access spectrum available and is typically less than 100 Mbps. Given that the cost to deploy a fiber-based solution is almost independent of the bandwidth provided, the cost per Mbps is higher at these typical base station bandwidths. This makes a fiber-based solution better suited to the aggregation layer of the backhaul network than the first tier.

Microwave solutions also have a fixed upfront cost, however, some solutions such as DragonWave's IP-based radios offer flexible bandwidth pricing to mitigate this. Microwave solutions also have simpler installations - eliminating the need for trenching fiber laterals, reducing cost and time to deployment. They do not scale to the same bandwidth as fiber-based solutions, however, it is quite possible to deliver several hundred Mbps per link using commercially available systems.

The cost-per-bit versus bandwidth is shown for fiber-based and microwave-based solutions in the graph below. A five-year NPV calculation has been used to derive an equivalent lease cost for the microwave link. This includes equipment, installation, antenna lease and annual maintenance costs. The fiber-based Ethernet is a market value assumption for a leased Ethernet service and includes an amortization of $20,000 per site for the fiber lateral build. (i.e. a few hundred feet - longer laterals can increase this build cost significantly)

As can be seen, the microwave solution is much more cost-effective than the fiber based solution in the 10 Mbps to 100 Mbps range typical of a single base station. Both solutions are lower than the range of $150 to $300 per Mbps per month for leased T1 circuits in North America and much lower than the higher leased circuit prices typical in most European countries. Thus, microwave is likely to be the technology used to connect access links to aggregation sites. Interconnection of aggregation sites will be a combination of microwave and/or fiber, depending on network size and availability of fiber-based assets.


Given this architecture and technology selection the distribution of total network spends in a WiMAX network can be determined. This is shown in the figure below and assumes an average of 100 CPEs per base station in 2006 increasing to 150 CPEs per base station in 2012, with an average cost ranging from $300 in 2006 down to $100 in 2012 Even with an optimized backhaul solution costs are still comparable with base station costs. This ratio stays -constant over the period 2006 to 2012 even with the assumed growth of backhaul bandwidth from 10 Mbps to 100 Mbps per base station due to the inherent scalability of the proposed architecture.

While RAN technology and hardware are clearly essential to the -delivery of personal broadband services, the profitability of the network is equally driven by backhaul network architecture and technology choices. It is imperative for network operators to pay attention to all in order to succeed in WiMAX deployments.

Alan Solheim is Vice President of Product Management for DragonWave Networks.

DragonWave Inc. designs, develops, markets and sells carrier-grade microwave equipment offering high capacity broadband wireless systems for network operators and service providers worldwide. http://www.dragonwaveinc.com