 | Issue: | Asia-Pacific III 2007 |
| Article no.: | 6 | |
| Topic: | Unclogging backhaul – architecture is not just technology… | |
| Author: | Björn Krylander | |
| Title: | CEO | |
| Organisation: | Cambridge Broadband Networks | |
| PDF size: | 348KB |
About author
Bjorn Krylander is the CEO of Cambridge Broadband Networks, a wireless transmission equipment developer. Mr Krylander came to Cambridge from UbiNetics, a 3G test and IP company, where he was CEO until its sale. Prior to UbiNetics, Mr Krylander held a variety of positions at Ericsson Inc., including Vice President TDMA and CDMA Mobile Phones in the US and Vice President and General Manager of the company’s Home Communications Business Unit. Mr Krylander holds a Master of Science in Electronic Engineering from Lund University.
Article abstract
The use of mobile broadband is growing rapidly and will accelerate greatly in the coming years. To meet the voracious demand that is developing, mobile operators will have to build up their infrastructure-especially the backhaul connections between local base stations and the core network. The point-to-point architecture typical of so many of today’s backhaul networks will be expensive to expand. Point-multipoint architectures handle data traffic peaks more efficiently and cost effectively and are much cheaper to build and maintain.
Full Article
After ten years of hype about the possibility, the time for true mobile broadband Internet is finally here. Around the world, Internet usage and services are becoming more widespread, more varied and altogether more demanding for the cellular operators that offer them. This shift from basic mobile voice services is made possible by HSPA. According to the Global Mobile Suppliers Association, GSA, 128 operators in 63 countries have commercially launched HSDPA services and a further 50 operators have committed to do so. The Asia-Pacific market already comprises more than half of the world’s current two billion wireless subscribers, according to industry analyst ABI Research. By 2011, when world usage rises to 3.5 billion, the Asia-Pacific region will have more than four times the number of wireless subscribers than the next largest region, Western Europe. For the end user, whether business professional or consumer, HSPA means true broadband on the move, supporting a variety of applications, including email, Web browsing, virtual private networks, video on demand and on-line enterprise applications. Operators are quoting huge data subscriber growth rates, such as a 20-fold increase in just six months since the launch of ‘all you can eat’ price plans. As data, rather than voice, becomes the driving force for growth and revenues, operators face major challenges ensuring their network infrastructure enables them to keep pace with demand. As well as coping with the rapid increase in bandwidth requirements, they have to manage the fact that data services exhibit entirely different traffic patterns than voice services, which are consistent and predicable. Additionally, since ARPU, average revenue per user, is not rising in parallel with the necessary increases in bandwidth, operators need to manage their costs carefully for both capital and operational expenditures. So, how is the voice-to-data paradigm shift in traffic usage supported? On the access side, the point at which services are delivered to the end user, a veritable alphabet soup of technologies – LTE, WiMAX, 3G, 4G, W-CDMA, HSDPA, HSUPA, UMTS, EV-DO – are in use. Vendors are falling over themselves to provide operators and end users with their best technology for the job. On the end-user side, according to the GSA, there are now 311 HSPA devices available, from 71 suppliers. The future for the access side of the network is clear; peak data rates will continue to increase (we will return soon to why the word ‘peak’ is so important). HSPA peak data rates are set to increase to 3.6, 7.2, 10, 14, 28, 42 and eventually 80Mbps, all within a single 5MHz W-CDMA channel. However, it’s not just about access. On the transmission side of the network, operators need to backhaul traffic from an increasing number of cellular base stations to the core network. This is typically achieved through point-to-point microwave links, with a capacity of 1 or 2 E1s/T1s per cellular base station sufficing until now to cope with even the peak traffic requirement from a cellular base station. As the peak data capacity needed by each cellular base station continues to grow, a point-to-point network, where by definition each cellular base station requires a dedicated link, becomes untenable. Each link has to be provisioned to cope with the peak data rate – the worst-case scenario – of its cell site, which is prohibitively expensive. The ratio of the peak to mean bandwidth requirement (i.e. the difference between the maximum amount of traffic a cellular base station needs to carry at a single point in time compared to the average amount of traffic it usually carries) is so large that the majority of the backhaul capacity is not required for most of the time, which is hugely wasteful. The reason for this peak-to-mean variation is because data applications, unlike voice applications where all users generate similar amounts of traffic, are ‘statistically violent’, or ‘bursty’, with music, email, video and enterprise applications all generating different types of data traffic with bandwidth needs. As peak data rates continue to grow, the peak-to-mean bandwidth variation will only continue to increase, further compounding the problem and increasing the required over-provisioning on the backhaul side of the network. Even if operators could simply throw an increased number of higher capacity point-to-point radios at the backhaul problem, apart from the expense, there is a huge environmental and community impact because of the sheer number of radio antennas required. Trying to counteract the upgrade problem by reducing the backhaul capacity needed by each cellular base station, by spreading the network over a greater number of smaller cells, makes the antenna problem of point-to-point radio links even worse. So how can operators overcome the technical problems of next-generation backhaul without compromising their business case by incurring costs so high that they cannot make a profit from their mobile Internet services? The solution is to use an alternative architecture for backhaul, leveraging the fact that the peak-to-mean ratio per cellular base station is ever increasing. Rather than using point-to-point links and having to provision for the peak data rate that each cellular base station may require, a point-to-multipoint solution overcomes these problems by utilising advanced traffic management techniques. This takes advantage of the relationship between the peak and mean data requirements of each cellular base station, and the advantage becomes even more apparent as the peak-to-mean ratio continues to grow. The fundamental difference between a point-to-point network and a point-to-multipoint network is the fact that a point-to-point network is comprised of a large number of independent, separately provisioned links, whereas in a point-to-multipoint network a single antenna at a central hub site connects to multiple cellular base stations. Using a carrier class point-to-multipoint architecture to overcome the ever-increasing backhaul bottleneck provides advantages both to operators and to the environment as a whole. The major impact for operators is a substantial reduction of the capital and operational expenditures needed to roll out advanced Internet services over their network; a point-to-multipoint architecture lets them expand without incurring crippling backhaul infrastructure costs that undermine their business case. The fundamental principle of a point-to-multipoint solution is that the capacity of a single central hub site can be shared between multiple cellular base stations by using advanced traffic management and statistical multiplexing. This means that rather than having to provision each link to cope with the maximum possible data throughput (the peak) for each cellular base station, the point-to-multipoint network simply needs to be provisioned so that the central hub station can cope with the sum of the average data throughputs from each cellular base station (the mean), and the traffic management software does the rest. As the peak-to-mean ratio becomes ever larger, the advantage of point-to-multipoint becomes more and more significant, substantially reducing the overall backhaul capacity an operator needs to provision, thereby resulting in a substantial reduction in costs. Simply reducing the number of antennas that need to be deployed substantially reduces capital expenditure. To provide backhaul capacity for ‘n’ cellular base stations, a point-to-point system needs 2 x n antennas, one at each end of each link. In contrast, a point-to-multipoint system requires n + 1 antennas, one situated at each cellular base station and one at the central hub site. [With a 5 base station system, for example, a point-to point architecture requires ten antennas and associated equipment, whereas a point-multipoint network requires just six antennas – a 40 per cent reduction.] Reducing the number of antennas also has a huge impact on operational costs, by reducing roof rental costs and the costs for site visits and maintenance. If an operator wants to increase the backhaul capacity as the network grows, it only needs to increase the central backhaul hub capacity, there is no need to replace multiple point-to-point links with higher capacity links. Additionally, when a new cellular base station is deployed, provisioning backhaul capacity to that site requires the installation of just one new antenna (the other end is already at the existing central hub site) rather than two antennas, further reducing opex. The ease of provisioning characteristic of point-to-multipoint and the fact that point-to-multipoint spectrum is typically allocated for a specific geographical area reduces the operator’s dependence upon third parties. This reduces dependence upon the regulator; because once the spectrum has been allocated the operator is free to use it without having to apply on a case-by-case basis for point-to-point spectrum licences. If an operator currently uses leased lines for backhaul, then the dependence upon third parties is reduced even more by adopting a point-to-multipoint wireless solution. In addition to the business impact upon individual operators, it is important to remember that the choice of point-to-multipoint vs. point-to-point is more than simply a technology choice. From the environmental perspective, just the reduction in the number of antennas on rooftops makes a huge positive impact on the local community. Additionally, the traffic management techniques enabled by a point-to-multipoint approach mean that maximum use can be made of the scarce natural resource, the radio spectrum. Finally, the choice of backhaul architecture is also about the need for operators to invest in a future-proof infrastructure that will enable them to continue to expand their business, offering an increasing variety of high bandwidth Internet services to their customers. This is particularly relevant in Asia-Pacific, where the number of mobile subscribers continues to grow more quickly than in any other region, and should reach nearly two billion by 2011. A carefully chosen point-to-multipoint solution will provide operators with a highly scalable platform that will support whatever portfolio of traffic and services the operator needs to offer, now and in the future.
