2.3 Drivers for the MBH Network Change

Higher and More Bursty Transport Load

Major traffic load for the 2nd generation mobile networks was, and is, voice – and that is often carried in a low bit rate format. For example, in GSM networks only 16 kbit/s is typically used per voice channel between the base station and the controller sites. Thus transport capacity required per base station site in these networks is very modest, in rural areas often only a fraction of a 2 Mbit/s link, and also in many city areas just a few 2 Mbit/s lines are needed (n*2Mbit/s, n = 1...4). In this type of network it is often possible to build such a high backhaul transport capacity that it is possible to carry all the traffic base stations are able to provide.

However, backhaul capacity needs are changing dramatically, and has already changed a lot: mobile data is growing fast and is already the major application in many 3rd generation mobile networks. And the growth of data traffic is just accelerating (see Table 2.1). This is supported by the further increase of mobile network capacities with new technologies and network generations, such as HSPA and HSPA+ and LTE.

Table 2.1 Examples of mobile data growth in 2010 (Source: Cisco VNI, 2011).

Increasingly, this new capacity is used for data intensive services and for video (especially where flat tariffs are applied), and the role of laptop PCs and new types of terminals is important here (see Figure 2.5). Additionally, often the instantaneous peak rates increase much more than the average traffic volume.

Figure 2.5 Example of mobile traffic forecasts (Source: Cisco VNI, 2011).

All this means that much greater total traffic will be present also in the backhaul links. Thus the transmission capacity required in the backhaul links for a base station site, especially in urban environments, is increasing strongly, and the actual load is much burstier than earlier. More and more often the MBH throughput will be critical for the actual mobile service capabilities, and for end-user experienced mobile service quality. If the backhaul capacities are not properly dimensioned initially, and later expanded in time, the backhaul network can become a bottleneck for the service quality and even for its reliable delivery.

More Cells Sites

Often there is a need, in addition to increasing capacities of individual cells, to increase the number of cells to accommodate the fast growing traffic, especially in city centers, business districts and other high traffic areas (hot spots). These smaller urban cells with high traffic potential need short distance high capacity transport links, with high flexibility in network building in these urban environments. Also lower backhaul costs than for earlier ‘big cell sites’ are now required (see next points).

Revenue Per Bit Decreasing

Another trend, happening at the same time as traffic grows, is the reduced revenue per bit of the mobile services. While many new services and traffic types mean ten or hundred times higher data volumes, end user ability and willingness to pay more for these new services is limited. The worst case is with the above mentioned flat rate services where even a strong traffic increase does not create related additional revenues. Thus even if the operator revenues for all the services increase, revenue per delivered bit can be just one tenth or even 1/100 of what it used to be. Thus cost efficiency is essential in the backhaul transport solutions and this is clearly becoming more and more important.

Lower Operational Costs

A third trend, related to the overall cost efficiency, is the need to operate all networks with smaller costs and more optimized organizations. In mobile backhaul networks this puts strong requirements for network simplification and automation of network operations to the widest extent possible. For example, self-healing properties of the networks reduce the need for immediate actions and thus contribute to reducing network maintenance costs (in addition to improving the service quality).

Network simplification in turn requires, for example, that the number of different technologies used within the network is reduced – as wide a use as possible of similar technologies is preferable. This creates scale benefits in equipment purchasing and savings especially in the operation and maintenance of the network, e.g. type of different skills needed is reduced. Also the effort needed for network planning is reduced when there are fewer different technologies and network layers.

Developments in General Transport

A significant change driver is also the development going on in general transport networks. The costs of packet-based transport solutions are now clearly lower than those of similar capacities with TDM or ATM technologies, and often power consumption of new packet-based equipment is also significantly lower than that of legacy equipment for similar capacities. In addition, the R&D efforts in the industry are focused on packet-based equipment, and their technical performance, power-efficiency and cost-efficiency continue to improve. Thus there is a strong need to benefit from these developments in the mobile backhaul solutions as well.

The developments of general transport networks also mean that leased line offerings are changing: in the longer term packet-based leased connections will have much lower price tags and more offerings and later also better geographical coverage than conventional leased lines, especially in case of higher capacity connections.

2.3.1 Mobile Service Developments and Traffic Growth

2.3.1.1 Traffic Forecasts Needed for Proper MBH Design

Mobile services developments drive traffic changes and determine what also needs to be carried over the MBH network. Evolution in the use of various mobile services has been extensively discussed in the literature, and is a hot topic in many research reports and newsletters. Several types of forecasts on expected mobile services usage and on expected revenues per service are presented. These kinds of forecasts are obviously essential from the mobile operator business case point of view.

However, from the MBH network design point of view, detailed distribution of mobile network usage for various types of mobile services is not so fundamental. Instead the total amount of traffic expected from each cell is essential, as well as the distribution of this traffic among the major traffic classes (e.g. voice/data, real time/non-real time, delay critical/not delay sensitive etc). This information is fundamental for the dimensioning of the MBH network, and it is also very important for some of the technical requirements of MBH, like for the delays allowed within a MBH network.

Therefore, even if exact service forecasts are usually not needed for the MBH network planning, total mobile traffic forecasts are essential, as well as forecasts for a coarse division of this traffic into various traffic classes. This is an input that is necessarily needed for an economic design of the MBH network; building a whole MBH network according to the maximum capabilities of new mobile systems is very rarely economically justified, as in many areas and in many cell sites real mobile traffic will only slowly (if ever) reach those mobile systems technical limits. Obviously there are also cell sites where these limits will be reached much more quickly; therefore mobile traffic forecasts should be available separately for different type of areas. And it can be emphasized that these traffic forecasts should be exactly for the mobile operator and mobile network for which the MBH network will be designed – more general traffic forecasts made, e.g. for a country, do not necessarily lead to an optimal MBH design.

It shall, however, be also noted that during a forecast period quite rapid changes can happen, often increasing expected traffic volumes. This can happen locally, caused by some new buildings or shops or service points within the base station coverage area, or more network wide due to some new services that get rapid popularity. This requires some margins in the backhaul dimensioning or flexibility for rapid upgrades, whichever is more economic in each case.

2.3.1.2 Traffic Peak Rates

The peak bit rates created or needed by various mobile services are obviously very important for the MBH design, as these rates must be supported also in all parts of a MBH connection – otherwise in reality those rates will not be available for the end-user. Many mobile services can adapt to the (transport) bit rate available, and can work over slower or faster connections, but clearly with different end-user experience; for example, many applications use TCP family protocols and increase bit rate until the network limit is reached. Certain mobile services are also adaptive but need a minimum bit rate to work well, e.g. many video services. And then there are also fixed bit rate services, e.g. video distribution to several users simultaneously. In all cases the whole MBH connection needs to support at least the single end-user peak bit rate.

A question about these service peak rates is very often a commercial question – due to competitive reasons, mobile operators want to promise certain available peak bit rates. Then this promised peak bit rate becomes the network design target instead of actual service related requirements or forecasts. From the MBH network point of view, it can have very significant cost implications whether the same peak rate is promised over all the mobile network area or not; for example, if it is promised also in all suburban and rural areas, all the MBH links in these areas must have a minimum capacity greater than the promised peak rate.

2.3.1.3 Average Service Bit Rate

Usage of broadband services is crucial from the total mobile network traffic volume point of view. Most important high volume service in many mobile networks is simple broadband access to internet; that can generate very high traffic volumes. When there are no fees based on usage (i.e. flat fee tariff) or other contractual or technical limitations, monthly cumulative traffic volumes can become very big in the case of ‘heavy net users’.

However, from the MBH point of view, most important is the average loading created during a typical ‘busy hour’, i.e. when a high number of users are simultaneously actively using their network access. Such a number can be much more difficult to forecast than the peak rates discussed above, as it depends on the expected number of simultaneous (broadband) users in a cell and on how coincident their traffic actually is. In the case of very bursty traffic (e.g. typical access to web pages) the probability of simultaneous peak loads is relatively low, but in the case of more continuous bit streams (e.g. large file downloads or video services) the probability of simultaneous high bit rates becomes rather high.

In spite of the difficulties in forecasting a ‘busy hour’ average load or an average bit rate of all users in a cell, at least some coarse forecasts should be attempted. Building a MBH network based on the maximum average bit rates which a mobile system can support is usually unnecessarily expensive – this particularly applies in the early phases of new mobile networks able to support high instantaneous bit rates (e.g. in early phases of a LTE network).

2.3.1.4 Traffic Distribution Into Classes

Distribution of traffic into classes with different requirements is also important, as the real time services are more demanding on connection capacity than other services; constant bit rate services are especially demanding. For the MBH design quite coarse distribution is satisfactory, mainly an estimate is needed about the share or volume of non-adaptive services; an example of a traffic forecast made for MBH planning is shown in Figure 2.6. These, especially some types of video services, need to be taken into account with a higher average loading than the more flexible services.

Figure 2.6 Example of mobile traffic forecast made for MBH planning.

2.3.2 Mobile Network Developments

Mobile service developments discussed above significantly change the traffic that needs to be carried over the backhaul network. In addition, there are mobile network changes, especially architectural changes (described more in detail in Chapter 03), which influence the backhaul network tasks and requirements. Some of the mobile network developments which are more important from the MBH network point of view are briefly discussed here.

2.3.2.1 More Native Packet Traffic

Total traffic volume expected from a base station site is strongly increasing with new mobile services, as discussed earlier, especially in urban base station sites. In addition, increasing portions of this traffic will be natively packet-based traffic, as new broadband services are all packet based. From the MBH point of view, packet traffic is also increasing because the new base stations have packet interfaces either for the majority of traffic (data traffic) or for all traffic (e.g. LTE base stations).

2.3.2.2 Flatter Mobile Network Architecture

New mobile networks, e.g. LTE, will have different mobile network architecture: there are no more controllers between the base stations and core elements. Therefore the transport domain becomes more unified, logical connections are directly from the base station to the core, and there will be more concentration points and other network nodes based on transport equipment.

On the other hand, some mobile core network elements may be distributed, placed closer to the base stations in the physical network, to increase mobile network capacity and to reduce delays. In such cases different logical connections may end at different sites, making transport configuration more complex.

2.3.2.3 Base Stations May be Connected to Several Core Sites

Another change made possible in the new architecture is that base stations may be connected to different core nodes and sites for redundancy reasons, even more than two different sites. Logical connections are then arranged from the base station site to several core sites, and traffic may move between those connections quite quickly, or even suddenly in the case of a link or a node failure.

2.3.2.4 Direct Links Between New Base Stations

Still a relatively new character having significant backhaul transport influence is that in the new architectures there are also logical connections directly between the base stations (e.g. in LTE architecture so called X2 interface, see Figure 2.7). In the transport domain such links can be implemented in different ways, either just logical paths using the same physical structure as before, or by adding new ‘transverse’ links to shorten the length of such connections. However, there are strong economic limitations to adding too many additional physical links into a backhaul network, and the practical solution is a compromise between ‘traditional’ tree and chain topology (often already built topology) and a more meshed topology (more direct physical links).

Figure 2.7 An example of mobile network architecture change, 3G WCDMA architecture vs. new LTE architecture (Note: new X2 interfaces between the base stations).

2.3.2.5 More Urban (Hot Spot) Cell Sites

In addition to increasing capacities of individual cells, there is a need to increase the number of cells to accommodate the fast growing traffic, especially in city centers, business districts and other high traffic areas (hot spots). These smaller urban cells with high traffic potential need high capacity short distance transport links. Also lower backhaul costs than for earlier ‘big cells’ are required.

2.3.2.6 Connecting Temporary Cells

Special events, like mass sport competitions or big outdoor music events, are likely to concentrate a lot of people into a small area and many of them expect to have their normal net connections during the event, especially during pauses. This may require arrangement of additional cells to increase the mobile network capacity for the duration of the event. Temporary cells may also be required after different kinds of catastrophes (e.g. earthquakes and floods).

These temporary cells need to have a connection into the regular network, i.e. to be connected to the permanent MBH network at a suitable node point. This in turn means that the backhaul network needs to have enough flexibility to make such connections possible and also that it must be capable of carrying the additional traffic offered.

2.3.3 Backhaul Cost-Efficiency Improvements

Both the service developments (influencing traffic volumes and types) and the mobile network developments (architectural changes) put their requirements on the MBH network design and implementation. In addition, there is a strong need to reduce the costs of backhaul networks, both relating to the investment costs for a certain capacity as well as considering the operational costs of the whole backhaul network.

2.3.3.1 Lower Costs for Higher Backhaul Capacities

High capacities required for the backhaul connections, especially in heavy traffic areas, and the need to keep network investments under control are together a big challenge for the design and dimensioning of backhaul networks.

Packet-based transport equipment usually has significantly lower costs for a certain transport capacity, i.e. lower costs per bit/s. This applies both for optical transport and microwave transport and even more to separate ‘traffic concentration’ nodes needed in the backhaul networks (e.g. packet switching equipment vs. TDM cross-connect nodes). This lower-cost-per-bit of packet-based transport equipment partly helps in avoiding a situation where network costs linearly follow the capacity requirements.

Another method is to try to increase sharing of physical links, either within the operator's own network or by sharing physical transport with another operator(s). A transport link of, say 1 Gbit/s capacity, is significantly cheaper than two separate links of half the capacity. Sharing within the network can be influenced by the transport topology design; sharing with other networks is obviously a matter of finding similar transport needs and then negotiations and agreements. Sharing can be especially useful in access links where a small number of cells do not necessarily use the link capacity efficiently but it must be dimensioned according to the peak capacity.

2.3.3.2 Better Ability to Handle Highly Bursty Traffic

The increasing peak-to-average ratio of the user traffic is also a challenge for an economic implementation of backhaul transport connections. The high peak rates easily speak for correspondingly high transport capacities; however, traffic burstiness can also be utilized for transport efficiency improvements in packet networks where there are no hard capacity allocations per user. The shorter the peaks are, the less is the probability of simultaneous demands, and thus a higher amount of statistical multiplexing can be assumed. Possible statistical gains also increase when the number of expected end users of a link increases – thus in a backhaul network higher statistical gains can be assumed above first traffic aggregation points.

2.3.3.3 Optimized Backhaul Dimensioning

Two previous points together mean that an effective tool to combat investment cost escalation is a good backhaul design and dimensioning strategy: the backhaul network capabilities and capacities are developed according to real service and business-based needs and not so much according to the theoretical maximum throughputs of the base stations.

In practice several backhaul network dimensioning approaches are possible, but the cost efficient ones often combine the ability to provide the promised bit rate to any single user and the expected average traffic of all users in the cells; the principle is shown in Figure 2.8. So the backhaul link capacities are not dimensioned according to the summed up end-user capacities, but based on expected usage profiles and expected number of users per link. Some, and often significant, statistical gain can be assumed already for the first link of the base station site, as there are typically several cells served from the same site.

Figure 2.8 Backhaul dimensioning principle (in the access tier).

2.3.4 Lower Operational Costs

Network operational costs are easily as significant for the mobile operator business case as the investments needed to create the network, and this applies also for the MBH networks.

Operational costs are related to network site running costs (e.g. equipment power consumption), network maintenance and fault repairs, to network re-configurations (when connecting new links and nodes and for traffic redistribution) and also to network planning and administration. All of these depend on the MBH network size and its complexity, and how its administration, planning and operation and maintenance (O&M) are organized. Even more, the relative amount of operational costs depends on the use of own facilities or leased connections and outsourced transport services – all lease fees are typically counted as network operational costs.¹

In general the MBH network share of a mobile operator's total operational costs depends very much on the area, size and density of the mobile network. Maintenance costs depend more than other costs on the geographical area covered, as the maintenance is more expensive in sparsely populated areas, areas with long distances between the network nodes and from the maintenance centers and manned service points. Transport's share of the operational costs can be kept under control by aiming at as simplified MBH network as possible and by using network automation on a wide scale.

2.3.4.1 Network Simplification

Network simplification means that the target MBH network has a simple clear topology, that is easy to understand and that helps in localization of faults and performance problems. Clear topology is important both in the physical network and in the logical network (i.e. in configuration of transport paths). For example, alternative paths between the network nodes are used only where connections are important enough and reliability requires redundancy. From the operational point of view also the use of all kinds of transport overlay structures should be avoided as much as feasible, and limited mainly to transition periods in the network. Overlay structures easily increase the effort needed for network maintenance, connection configurations and network planning, and overlay structure using different technologies requires keeping competences to manage all of them. In addition, personnel capable and skilled in the maintenance of the older generation equipment are becoming with time more scarce, and costs are increased in finding and keeping such personnel in house.

2.3.4.2 Similar Technologies

Use of similar technologies and similar kind of equipment in the MBH network clearly contributes favorably to network simplification. Thus from the operational point of view the target network should contain as few different technologies as possible, and often a fully packet-based MBH network is a reasonable target, at least in the longer term. Therefore, removing some network layers can significantly reduce operational effort and thus costs (e.g. removing the ATM layer used for WCDMA traffic, and/or reducing use of TDM links when the network is upgraded).

Also, within packet-based transport technologies a mobile operator may benefit by creating a strategy for focusing on the use of only two or three types of packet technologies as widely as possible, especially in the MBH network parts built by the mobile operator itself. Even further simplification is possible by using just a few types of equipment in the MBH network, and the network O&M may benefit a lot from this kind of simplification. However, such an approach needs to be carefully judged against possible trade-offs in future equipment purchasing (it may limit vendor competition in the following network expansion phases).

2.3.4.3 New Equipment with Better Performance and Management

Newer generation equipment generally has better power-efficiency, i.e. it uses less power for a similar amount of data, and this obviously reduces the bill to be paid for the electricity; and when a site requires air conditioning for keeping internal temperatures within an operating range of equipment, there is double benefit, as lower power consumption also means less heat production.

Newer equipment also has generally better and more extensive remote management capabilities so that site visits to perform some operations locally are less likely – fewer site visits are obviously a big contributor to lower maintenance costs.

2.3.4.4 Network Automation

Network wide automation is another way of reducing operational costs. Good network O&M tools with automated routine functions increase productivity of the network management teams, as less manual work is needed in the most common tasks and effort can be focused more on network optimization. For example, creation of new logical connections within the MBH network, or modification of existing ones, can be carried out with little manual work.

Also, automation within the MBH network itself reduces costs; for example, effective protection of connections (automatic switching to alternative paths) gives more time for maintenance teams, and reduces the likelihood that expensive night or weekend repairs are needed. It is worth noting, however, that extensive automation within the network nodes can in some cases mean trade-off with the network simplification, so that judgment is needed in priority of the goals.

2.3.5 Developments in General Transport

Last but not at all least, a significant change driver is developments going on in general in transport networks. The costs of packet-based transport solutions have become much lower than those of similar capacities with TDM (or ATM) technologies, as the R&D efforts of the industry have been concentrated already for some time on packet-based technologies. And more recently also the sales volumes of the packet-based transport equipment has significantly increased, meaning volume and scale benefits in their production and distribution. At the same time, technical performance of the packet-based transport solutions has been significantly improved, including the power consumption (power per bit), and equipment performance.

Thus, there is a strong push to benefit from these developments in the mobile backhaul solutions as well. Packet-based technologies have proved their superiority first in the MBH backbone networks where the traffic volumes are the highest, but today their superiority is clear also in aggregation networks. This development has continued and nowadays also affects MBH access networks where packet technologies offer higher cost-efficiency for high volume and increasingly bursty mobile traffic.

The development of general transport networks also means that leasing offerings are changing: packet-based connections will be much more widely available and will have lower price tags. There will also be available packet connections of different quality classes and more or less guaranteed bandwidths or throughputs. On the other hand, these network parts are cost optimized for fixed traffic, and technology and feature selections made for that purpose. Thus any mobile specific requirements typically mean increased price for the leased connections, or that such connections will be more difficult to obtain.