More spectrum, high spectral efficiency and small cells will provide up to 1000 times more capacity in wireless access. In the world of wireless, Shannon’s law is the one fundamental rule that defines the physical limits for the amount of data that can be transferred across a single wireless link. It says that the capacity is determined by the available bandwidth and the signal to noise ratio – which in a cellular system typically is constrained by the interference.
Therefore the first lever to increase the capacity will be to simply utilize more spectrum for mobile broadband. In total the entire spectrum demanded for mobile broadband amounts to more than 1,100 MHz and a large amount (about 500 MHz) of unlicensed spectrum at 2.4 GHz and 5 GHz can provide additional capacities for mobile data. Of course reaching an agreement on spectrum usage requires significant alignment efforts by the industry and is a rather time consuming process. Therefore it is also necessary to look at complementary approaches such as the Authorized Shared Access (ASA) licensing model, which allows fast and flexible sharing of underutilized spectrum that is currently assigned to other spectrum-holders such as broadcasters, public safety, defence or aeronautical.
A key challenge associated with more spectrum is to enable base stations and devices to utilize this larger and a potentially fragmented spectrum. Here technologies such as intra- and inter-band Carrier Aggregation will be essential to make efficient use of a fragmented spectrum.
The second lever for more capacity will be to address the interference part of Shannon’s equation. This can be achieved for example through beam forming techniques, which concentrate the transmit power into smaller spatial regions. A combination of multiple spatial paths through Coordinated Multipoint Transmissions (CoMP) can further increase the capacities available to individual users. We believe that with the sum of these techniques the spectral efficiency of the system can be increased by up to 10 times compared to HSPA today.
Advanced technologies and more spectrum will help to grow capacity by upgrading existing macro sites for still some time. However, a point will be reached when macro upgrades reach their limits. By 2020 we believe mobile networks will consist of up to 10…100x more cells, forming a heterogeneous network of Macro, Micro, Pico and Femto cells. Part of this will also be non-cellular technologies such as Wi-Fi, which need to be seamlessly integrated with cellular technologies for an optimal user experience.
Although the industry today has not defined what 5G will look like and the discussions about this are just starting, we believe that flexible spectrum usage, more base stations and high spectral efficiency will be key cornerstones.
The capacity demand and multitude of deployment scenarios for heterogeneous radio access networks will make the mobile backhaul key to network evolution in the next decade. The backhaul requirements for future base stations will easily exceed the practical limits of copper lines. Therefore from a pure technology perspective, fiber seems to be the solution of choice. It provides virtually unlimited bandwidth and can be used to connect macro cells in rural areas and some of the small cells in urban areas. However the high deployment costs will prevent dedicated fiber deployments just to connect base stations in many cases. Due to the number of deployment scenarios for small cells, from outdoor lamp post type installations to indoor, we believe a wide range of wireless backhaul options will coexist including microwave links and point to multipoint link, millimetre wave backhaul technologies. For many small cell deployment scenarios (e.g. for installations below rooftop level) a non-line-of-sight backhaul will be needed. The main options here are to either utilize wireless technologies in the spectrum below 7 GHz or to establish meshed topologies.
Besides pure network capacity, the user experience for many data applications depends heavily on the end-to-end network latency. For example, users expect a full web page to be loaded in less than 1000ms. As loading web pages typically involves multiple requests to multiple servers, this can translate to network latency requirements lower than 50ms. Real-time voice and video communication requires network latencies below 100ms and advanced apps like cloud gaming, tactile touch/response applications or remotely controlled vehicles can push latency requirements down to even single digit milliseconds.
The majority of mobile networks today show end-to-end latencies in the range of 200ms-500ms , mainly determined by slow and capacity limited radio access networks. Therefore the high bandwidth provided by future radio access technologies and the use of fast data processing and transmission will provide a major contribution to reduce the network latency. Due to the amount of data being transferred the user perceived latency can be much higher than the plain round-trip-time. Thinking of future ultra high resolution (UHD) real time video applications this clearly motivates the need for further technology evolution.
Equally important is the real traffic load along the end-to-end path in the network. A high traffic load leads to queuing of packets, which significantly delays their delivery. When attempting to solve this, it is not efficient to just overprovision bandwidth in all network domains. Instead latency sensitive media traffic might take a different path through the network or receive preferred treatment over plain data transfers. This needs to be supported by continuously managing latency as a network quality parameter to identify and improve the bottlenecks. In return, low latency traffic could be charged at a premium, providing network operators with new monetization opportunities.
One physical constraint for latency remainins: Distance and the speed of light. A user located in Europe accessing a server in the US will face a 50ms round-trip time due simply to the physical distance involved, no matter how fast and efficient the network is. As the speed of light is constant, the only way to improve this will be to reduce the distance between devices and the content and applications they are accessing. Many future applications such as cloud gaming depend on dynamically generated content that cannot be cached. Therefore the processing and storage for time critical services also needs to be moved closer to the edge of the network.
The introduction of additional radio access technologies, multiple cell layers and diverse backhaul options will increase complexity and bears the risk that network OPEX will rise substantially. This is why the Self- Optimizing-Network (SON) is so important. SON not only increases operational efficiency, but also improves the network experience through higher network quality, better coverage, capacity and reliability. Extending the SON principles now to a heterogeneous network environment is a challenge and opportunity at the same time.
Fortunately, big data analytics and artificial intelligence (AI) technologies have matured in recent years, mainly driven by the need to interpret the rapidly growing amount of digital data in the Internet. Applied to communication networks, they are a great foundation for analyzing Terabytes of raw network data and to propose meaningful actions. In combination with AI technologies, actionable insights can
be derived even in the case of incomplete data; for example machine-learning techniques can find patterns in large and noisy data sets. Knowledge representation schemes provide techniques for describing and storing the network’s knowledge base and reasoning techniques utilize this to propose decisions even with uncertain and incomplete information. Ultimately we believe that both, big data analytics and AI technologies will help to evolve SON into what we call a “Cognitive Network”, one that is able to handle complex end-to-end optimization tasks autonomously and in real time.
Customer Experience Management (CEM) can provide insights that will enable operators to optimize the balance of customer experience, revenues and network utilization. Cognitive Networks will help to increase the automation of CEM enabling network performance metrics to be used to govern the insight/action control loop, as well as experience and business metrics. This again increases the operational efficiency and at the same will be the prerequisite to deliver a truly personalized network experience for every single user.
The big data analytics and AI technologies introduced with the Cognitive Networks will be the foundation for advanced customer experience metrics. The ability to deal with arbitrary amounts of data in real time will allow a much more detailed sensing of network conditions and the resulting user experience in real time.
It also will be the foundation for large-scale correlations with other data sources such as demographics, location data, social network data, weather conditions and more. This will add a completely new dimension to user experience insights.
Cloud technologies and being able to provide computing and storage resource on-demand have transformed the IT industry in the last years. Virtualization of computing and storage resources has enabled the sharing of resources and thus their overall efficiency. Virtual cloud resources can also be scaled up and down almost instantly in response to changing demand. This flexibility has created completely new business models. Instead of owning infrastructure or applications it is possible to obtain them on-demand from cloud service providers. So far this approach has mainly revolutionized IT datacenters. We believe that similar gains in efficiency and flexibility can be achieved when applying cloud technologies to Telco networks. Virtualization will allow decoupling of traditional vertically integrated network elements into hardware and software, creating network elements that consist just of applications on top of virtualized IT resources. The hardware will be standard IT hardware, hosted in datacentres and either owned by the network operator or sourced on-demand from third party cloud service providers. The network applications will run on top of these datacentres, leveraging the benefits of shared resources and flexible scaling.
Also user plane network elements such as base stations will be subject to this paradigm shift. Over time, the migration of network elements in combination with software defined networking will transform today’s networks into a fully software defined infrastructure that is highly efficient and flexible at the same time.
Efficient radio technologies, high utilization and network modernization will reduce the network energy consumption, another important cost factor for operators. Having the forecasted traffic growth in mind, reducing the network energy consumption must be a major objective. The focal point for improving network energy efficiency will be the radio access, which accounts for around 80% of all mobile network energy consumption. Ultimately the energy efficiency that can be achieved depends on the pace of network modernization. Efficiency gains materialize only when the new technologies are introduced into the live network. Determining the right pace for modernization requires careful balancing of CAPEX and OPEX. We believe that energy efficiency can beat the traffic growth – which makes keeping the network energy consumption at least flat a challenging – but achievable goal.
