Despite scaremongering about traffic growth and falling revenue, leading operators around the world have demonstrated that increases in mobile broadband subscriber numbers and revenues exceed traffic growth
Those operators have proved that revenue and traffic are not decoupled, and that growth in mobile broadband is profitable. The key areas of HSPA (High-Speed Packet Access) capacity challenges are traffic capacity of the radio interface, the transmission network and the core network together, and the network signaling capacity. The first concentrating node in WCDMA/HSPA networks, the Radio Network Controller (RNC) can each handle hundreds of thousands of mobile broadband subscribers, so there is actually no limitation in this part of the network, assuming reasonable quality of service (QoS) measures are in place to prevent excessive usage.
Real-life field data shows that traffic growth always follows subscriber growth. Higher up in the network, the control and packet-switching nodes and transmission resources that serve the radio access network will need to be dimensioned in line with traffic growth. The immediate challenge is to ensure broadband capacity is available through proper dimensioning, software upgrades and targeted capital network investment where it’s needed.
The vast majority of radio sites use less than five per cent of their capacity. Typically only four per cent of radio sites are more than 50 per cent utilised in the most developed mobile networks. In certain busy locations, targeted investment in radio capacity may be needed at sites that have reached 75 per cent average utilisation, but the issue is not in the radio interface.
This point can be illustrated by a simple calculation. Based on each mobile broadband subscriber using 1GB of data each month on average, operators can predict a busy-hour (BH) data usage of 5kbps per user in the uplink and downlink. A typical HSPA 5MHz carrier can deliver a total data capacity of 5Mbps.
In a three-sector HSPA radio base station, the total data capacity is 15Mbps – enough to serve 3100 BH users. It is important to remember that most HSPA operators have enough spectrum to deploy three carriers, meaning that capacity can, in most cases, be tripled by adding new radio units.
The HSPA radio interface has ample capacity to deal with significant growth in mobile broadband traffic volumes, as each one-carrier three-sector cell can deliver up to 15Mbps data. However, in some areas, growth in data traffic is putting a strain on the backhaul and transport networks, that carry the traffic between the radio network and the call control nodes which than requires operators to make investments to handle mobile broadband growth.
There are a number of options, including internet protocol (IP) over fiber, IP over microwave, business broadband Asymmetric Digital Subscriber Lines (ADSL) and Very high-speed Digital Subscriber Lines (VDSL). The best solution for backhaul and transport network expansion depends on local market service availability and pricing. Typically, the break-even point for leasing carrier-grade ethernet links is at around 3-4 E1 (each of 2Mbps), but this is generally possible only in metro areas.
Operators have the option of deploying business broadband connections for data offload during busy periods. They can also deploy their own high-capacity microwave links, which provide longer-term flexibility and cost control after the initial investment.
It has been shown that as much as 94 per cent of mobile data traffic is generated by as little as four per cent of users. If left unchecked in a well-loaded network, heavy data users will have a serious effect on network performance and other users’ experience.
As with the transmission network, core network capacity can be scaled in direct proportion to traffic volume or subscriber numbers. Core network nodes, such as the Serving General Packet Radio Services (GPRS) Support Node (SGSN) and Gateway GPRS Support Node (GGSN) are large, powerful routers where mobile data traffic is highly aggregated. Capacity planning is relatively straightforward and efficiency is improved further the larger these nodes become.
However, there are a number of additional features being rolled out for the core network that provide more efficient use of existing resources and remove the need for immediate additional capacity investment. One such feature is 3rd Generation (3G) Direct Tunnel, which provides a direct IP payload tunnel between the RNC and the GGSN, rather than having the payload routed via an SGSN. A related aspect of capacity management in the core network is QoS management, especially to control traffic from the heaviest users and applications on flat fee agreements.
The aim of 3G Direct Tunnel is to increase the capacity of the core network without the need for additional equipment. The actual data payload is shifted away from the SGSNs, which means they have more capacity to deal with signaling and mobility management. Field measurements show that 3G Direct Tunnel can enable capacity savings of up to 90 per cent for the SGSNs.
Increasing smartphone usage can present a challenge for operators if their networks have not been optimised for the devices. The challenge is not about data usage – which is only around 100MB to 500MB per month for a typical smartphone user, compared with five to 10 times the figure for mobile-broadband-enabled laptop users – but rather has to do with the way the signaling system has been configured.
The issue is that many existing implementations of HSPA do not have the Cell_PCH (Cell_Paging Channel) or URA_PCH states enabled, and as a consequence, smartphones are returned to the idle state frequently. This would be acceptable if smartphones only needed data access occasionally, but many popular smartphone make 10 to 100 requests each hour, especially if people are using web applications originally designed for fixed-broadband-connected PCs.
Measurements by Ericsson in live networks have shown that keeping devices in the URA_PCH state for 30 minutes – rather than allowing them to drop back to idle mode each time a data transfer is completed – reduces the channel connection establishment requests significantly.
Another way operators can reduce the number of state changes and the overall signaling load is to keep the number of uplink states for HSPA to a minimum. As all new smartphones are capable of using HSPA on the uplink, Ericsson recommends having just four uplink states, rather than the seven uplinks allowed as a standard. Doing this has been shown to reduce the number of uplink state changes and the signaling load by about 40 per cent.
Ericsson predicts that there will be more than 500mn PCs and 3bn smartphones worldwide by 2015. In addition, for every subscriber, there will be multiple SIM cards for additional products, such as e-book readers, navigation tools and portable multimedia devices. This will drive revenue growth for some time. If traffic per subscriber proves to be roughly constant, the traffic growth curve should follow the same path as the subscriber growth curve.
In reality, market experience shows that after the initial take-up of mobile broadband services by early adopters, more typical users start signing up to the services and begin to dominate the subscriber base. Based on current HSPA network utilisation – which is mostly below 10 per cent – the cost-per-gigabyte appears high, given the significant investments that have been made in HSPA networks to date. The proportion of actual costs allocated to mobile broadband rather than voice tends to be high too, as mobile broadband is much more efficient than voice. However, it is important to understand that the cost-per-gigabyte varies significantly with the number of subscribers using the network and the volume of data they are using.
Far from being an additional cost burden, network traffic growth is a result of subscriber growth and traffic growth per subscriber, and actually drives down operators’ cost-per-gigabyte. The cost of adding mobile broadband subscribers will continue to be mostly marginal, until a time when capital investment is needed at radio sites.
Through a combination of 64 QAM, multiple-input multiple-output (MIMO) transmission and multi-carrier (dual and quadro) technologies, HSPA operators have a well-defined evolution path to take their networks to high levels of data rate performance. Spectrum efficiency is increased from roughly 1bps per Hz to 1.5bps and 2bps per Hz. With data rates as high as 168Mbps and beyond available from HSPA in the next couple of years, combined with the significant increase in spectral efficiency, operators have a very cost-effective way of improving their mobile broadband offering while meeting future demands for higher capacity.
Despite the ongoing debate about their capacity to manage, WCDMA/HSPA operators do not need to invest in completely new radio technologies to offer their subscribers an ever-improving, high-speed mobile broadband experience. They can be confident they already have the radio network capacity to deal with substantially more mobile broadband subscribers and traffic in the future. With planned, targeted investment in expansion of transmission and control node capacity, mobile broadband traffic growth will be manageable. These costs are marginal and will be more than offset by subscriber revenue growth.
