Hundreds of Columns Sorted by Topic
WiMAX Won't Go Far Without Backhaul Scalability
by Alan Solheim, VP of Product Management
2/13/2008
The introduction of WiMAX appliances (whether mobile or fixed) at the edge of the network creates the opportunity to deliver exciting new services such as personal broadband, streaming video and interactive gaming. Although the WiMAX radio access network can deliver the required connectivity, we need to rethink the backhaul network or these new services will starve for bandwidth and the entire ecosystem will be endangered. The introduction of the iPhone provides a compelling example of what happens when a new service is introduced without the network being ready to deliver the required support. For early iPhone adopters, limited network bandwidth prevented all but the most rudimentary services, which took a painfully long time for each response.
So what are the requirements for the WiMAX backhaul architecture?
bandwidth scalability
improved economics
multi-service support
The first decision is to select the backhaul technology that will deliver on these requirements and then identify the features that are required to optimize the total network architecture.
The core network consisting of Dense Wavelength Division Multiplexing (DWDM) fiber links in general has ample capacity, however the connection between the base station and the fiber point of presence represents the bottleneck in today's network. This backhaul has been historically provided either by leased T1/E1 lines from the local wireline carriers or via low speed PDH radio links. North America has preferred the leased line approach due to favorable T1 pricing and Europe has preferred the microwave approach due to higher costs from the wire line service providers. This approach has two limitations. First it is a circuit switched network and so is not optimized for the next generation of IP based services. Dedicated circuits prevent the operator from taking advantage of the statistical multiplexing opportunity inherent in IP based services and preclude the simplified provisioning and supervision available from IP based networks. Second these historical solutions can not scale to deliver the required bandwidth. For the leased line case this is because the cost scales more or less linearly with the increasing bandwidth and for the PDH microwave case because the installed radios can not deliver the required bandwidth.
Alternative technologies such as DSL, fiber to the base station and IP based microwave radios can address some or all of these issues. For DSL technologies the bandwidth distance product means that they can only deliver the required bandwidth to base stations that are in close proximity to the central office. They are also prone to disruption by environmental effects such as lightning and other electromagnetic interference. Finally they require access to the copper circuits which are owned by the wire line carrier -- the very entity that the mobile operator is trying to wean itself off of. Fiber to the base station holds the promise of almost unlimited bandwidth, however even in the most developed countries the penetration of fiber to base stations is below 10%. As the adoption rates of the new services increase, the mobile operators will have to add additional base stations in order to provide the required bandwidth per user over a fixed amount of radio access spectrum. This will only increase the number of base stations that need to be connected. At construction costs of on the order of $100 per foot, and with increasing pressure from urban councils to minimize the disturbance from construction in urban centers the business case for fiber to the base station becomes unattractive for a majority of sites. Alternatively, licensed packet radios can provide the bandwidths required and the benefits of an IP based network with the network availability required by the operators and do not require extensive civil construction to be deployed. Although all three technologies will be used, whether in an owned or leased model it seems clear that in the near term a significant portion of the network will use IP radios for the backhaul solution. This is the technology that we will discuss in further detail.
It is essential to match the expenditure on network capacity with revenue from new subscribers. Truck rolls or changes to RF link licenses to add network capacity are cost-prohibitive. The answer is to add bandwidth via software. Several vendors allow the operator to add bandwidth with the stroke of a key in the Network Operation Center without anyone having to physically touch the backhaul links with promises that this meets the software scalability requirement identified above. If this is done via changing the modulation index, however, the RF characteristics will change, the link availability will degrade and in most jurisdictions the operator will have to apply for a change to the link license. In most cases this means that effectively the promise of in service scalability is not met. A better alternative is to separate the traffic policing function and the RF modulation. Doing this allows the RF modulation to be set at installation which determines the maximum available capacity without engineering changes. Then, using traffic policing functions common to many IP networks the throughput available to the operator can be controlled independently of the RF propagation. Software keys used to change to the permitted bandwidth then become analogous to adding additional timeslots in an SDH network and can truly be done in service. If this is coupled with a pricing model that ties purchase price to bandwidth keys the operator has the opportunity to match the network cost to the bandwidth demand (and hence the revenue from that bandwidth).
A second key feature for cost-effective bandwidth scaling is adaptive modulation. The conditions that limit the availability on a microwave link are RF fades due to rain. Since this only happens periodically, most of the time the link is capable of carrying much more capacity than (than What???). Adaptive modulation lets the Network Operator engineer each link to deliver the high priority traffic with the desired availability and then increase the modulation index to gain access to the full link bandwidth under normal conditions at virtually no additional cost. It should be noted that compared to license exempt radios, which must be constantly shifting modulation index to compensate for interference and multi-path reflections, the licensed radio links that we are discussing only use modulation shifting under infrequent bad weather. As a result hitless switching is not important and minimizing the switching instances provides for higher overall network throughput and stability.
Lifecycle cost management requires paying attention not only to the cost of capital expenses for the infrastructure, but also to the installation, operations and leasing costs. These costs can easily exceed the capital cost when considering a 5 year net present value of a backhaul network. The key here is integration to minimize the footprint of the backhaul solution, both indoor and outdoor.. This reduces the equipment cost and the tower lease cost, reduces the infrastructure modifications that are required to deploy the new network and simplifies the installation and troubleshooting of the network.
Finally, the network must support multiple types of services to maximize the utilization of the backhaul asset. Next generation services require flexible QoS, low latency and flexible bandwidth allocation. Traditional voice circuits require high quality T1 or E1 circuits. Since the bandwidth of the new services will dominate the total capacity requirements it makes more sense to use IP as the convergence technology. Using a combination of pseudowire technologies and high performance packet radios, operators can meet these needs without sacrificing the OPEX efficiencies of a converged packet based network. Alternatives which carry both the TDM and IP separately over the RF link do not achieve this convergence and force the operator to manage and maintain both the TDM and IP connections at each intermediate point in the network. This drives the need to deploy additional hardware to aggregate, provision and protect each type of traffic separately. Using a converged IP approach with pseudowire for the legacy traffic means that all pass-through traffic remains in the IP format, significantly reducing the cost at intermediate nodes and eliminating the need for high E1 count radios. Further, the pseudowire devices can aggregate E1s or T1s into higher rate interfaces at the hub site, further reducing the network costs. This solution also provides future proofing as well. As the traffic continues to migrate from TDM to IP interfaces nothing needs to be done to the network to accommodate this change in traffic weighting.
The final consideration is not a technology one, but one of organizational dynamics. Carriers who have historically leased the backhaul network do not have the expertise or manpower to build and operate these networks. Also the capital markets are opposed to the carriers adding more manpower. This results in many outsourcing solutions and opportunities for third party wholesale bandwidth providers. In many cases it takes more time and energy for carriers to sort out these organizational issues than to make the technology selections discussed above. However, for the new services to be successful these issues must be resolved and the carriers that are able to muster the will to do so will prosper, and those that stumble will be left behind.
About the Author
Dr. Alan Solheim holds a PhD in Electrical Engineering from University of Waterloo. With more than 21 years of industry experience in Telecommunications. Prior to joining DragonWave, Alan was Chief Technology Officer at Innovance Networks, a reconfigurable optical networking start-up. Prior to Innovance, Alan was a Vice President at Nortel Networks responsible for market strategy in the metro transport group. Alan has extensive experience, notably working on 6 generations of fiber optic transmission systems, and the system design authority for Nortel's OC-192 program. Alan has over 50 patents that he has been principal or co-author of, numerous papers in industry journals and has acted on a variety of industry and conference committees.
About DragonWave
DragonWave designs, markets and supports broadband, wireless networking products for service providers and enterprises requiring reliable, predictable, interference-free, high-bandwidth transmission of real-time, IP applications. DragonWave products meet the demands of a wide range of applications as well as delivering a value proposition that enables operators and service providers an "invest as you grow" capability that leverages profitable growth. DragonWave is headquartered in Ottawa, Canada's high-technology capital.
Send us your response to this article.
Learn How to Get Your Column Published on this Site
