Future 5G test methods will need to provide a high confidence to operators that the technology and services are implemented according to specification, and that the quality of service is matching to the requirements of the application or service being delivered.
Madhukar Tripathi, Anritsu India Pvt Ltd
A new mobile network “generation” usually refers to a completely new architecture, which has traditionally been identified by the radio access: Analog to TDMA (GSM) to CDMA/W-CDMA and finally to OFDMA (LTE). So the industry has started now to refer to the next fundamental step beyond fourth generation OFDMA (LTE) networks as being “5G”. It is clear that 5G will require a new radio access technology, and a new standard to address current subscriber demands that previous technologies cannot answer. However, 5G research is driven by current traffic trends and requires a complete network overhaul that cannot be achieved only through gradual evolution. Software-driven architectures, fluid networks that are extremely dense, higher frequency and wider spectrum, billions of devices and Gbps of capacity are a few of the requirements that cannot be achieved by LTE and LTE-Advanced.
The concept for “5G” is both an evolution of wireless networks to meet future demands for data, and a revolution in architecture to enable a flexible and cost efficient network that can be efficiently scaled. These are the network operator operational demands on the network and technology, but they are driven by the demands for the type user experience which should be offered. These user experience demands that provide the underlying requirement for “5G” are Capacity enhancement, Coverage enhancement, Convenience enhancement
5G development, as any other technology, can be scheduled into 3 main stages: firstly, the definition of requirements and key technologies, secondly the development and standardization processes, and eventually real demonstrations plus further commercial deployments.
The standardization process started during 2015, to lay out a plan of activities to review requirements and technologies and then produce a set of standards. This activity is taking the guidance from WRC15 in terms of candidate new frequency bands to study for 5G, and re-use of existing mobile broadband frequencies. The 3GPP schedule for development, writing and approval of standards may last until the second half of 2019, so in parallel we will see a huge effort in R&D against these emerging standards. We expect to see first deployments based on the partial Release 15 standards in mid-2018.
5G Requirements
There are several projects and groups working worldwide to define 5G needs, technology requirements, user requirements, and other views on what 5G should be. The mobile network operators have built up an ecosystem community called Next Generation Mobile Networks (NGMN) which is creating a requirements document from the perspective of wireless network operators. NGMN had previously agreed requirements for 4G networks, and selected LTE as a preferred technology. The NGMN is an operator led forum, but with participation from the equipment/network vendors and supply chain, to try and align the industry needs.
Technical Challenges and Targets
There are several key challenges to be met with future 5G networks, looking at the performance needs of future networks and identifying affordable technologies which may be developed to support these.
5G is promoting targets in all different service layers, it is an all-optimized network designed to support the highest user data rate with unlimited coverage and within challenging mobility scenarios, capable for a massive number of connections per km2 whilst also reducing the latency to the “immediate” feeling and the cost and power consumption of devices to their minimal possible figure. Ensuring all these parameters would transform 5G into the most complete wireless network ever seen, tremendously attractive for automotive, M2M and even upcoming futuristic applications. This is what we call “Performance oriented” use case.
Volume of data
Data will be one of the key drivers for 5G. Firstly, continuing the legacy from LTE, voice will be totally handled as an application simply using the data connectivity provided by the communication system and not as a dedicated service. This is an additional increase in a data growing pace of between 25% and 50% annually and is expected to continue towards 2030 due to the increase in size of content and the number of mobile applications requiring high data rates, the rise in screen resolution with the recent introduction of 4K (8K already in development and expected beyond 2020) and the developments in 3D video.
There is much discussion in the wireless communications industry about Machine to Machine (M2M) communication being a big driver for future growth, as an enabling transport technology for the Internet of Things (IoT). Although there are many case studies and potential business models being developed, there is one segment of the industry which is already pushing forward in this area, the automotive business.
5G will introduce more requirements of the transport network and overall architecture in order to ensure support for any-to-any communication, so not only device to device but also node to node self-backhauling. Intelligent Transport Systems (ITS) for traffic control and passenger safety purposes could also benefit from this kind of communication technology.
Research subjects
The academic community and industry research programs, are currently investigating a number of subjects to identify and develop key technologies. The research subjects are based around the generic technical needs for 5G to meet the objectives and targets previously outlined.
Key technologies
Based upon the research subject areas discussed, there are a number of specific technology needs which are currently being investigated to identify specific solutions and candidate technologies to be used within 5G networks.
Network design
A parallel evolutionary trend to 5G is software virtualization and cloud services, where the core network is implemented onto a distributed set of data centers that provide service agility, centralized control, and software upgrades. SDN, NFV, cloud, and open ecosystems are likely to be the foundations of 5G and there is an ongoing discussion about how to take advantage of these into new network architectures.
SDN, NFV, and cloud architectures will play a vital role in next generation of network architecture. It can provide an implementation model that easily and cost effectively scales with the volume of data and vast increase expected in connected devices. Secondly, it allows service providers to more easily and efficiently manage a dynamic network, with great flexibility to re-configure the network functions and capacity. Thirdly, SDN/NFV provides a practical implementation for more efficient routing of both User Plane and Control Plane data, to reduce total data volume travelling across the network, and to reduce latencies with more direct routing than an architecture based on centralized hardware functions. This will better support the separation of Control Plane and User Plane data, and the separation of downlink and uplink data.
New frequencies for radio access
As the RF spectrum up to 6 GHz has been fully allocated to different uses and services, the need for wider bandwidth of radio to support higher capacity data links is pushing the industry to investigate higher frequencies. Even if additional frequency bands below 6 GHz were made available for mobile communications, there are still fundamental limits on the amount of bandwidth that could possibly become available.
Mm-wave could be used by indoor small cells to provide very high-speed connectivity in confined areas. The high-frequency nature of mm-wave means antennas can be very small with only a small impact on device form factor. A number of market analysts still believe mm-wave is a radical technology and may require many years of R&D to be cost-effective for the mass market. There are still key issues with the performance of semiconductors at these frequencies, giving an impact on the cost and power consumption of devices such as amplifiers and mixers, which fundamentally limit the range/sensitivity of radio transceivers. This technology is now evolving from specialist high cost markets to becoming affordable for high volume and low cost massed markets as the semiconductor technology is further developed.
The combination of mm-Wave NLOS and mesh networking gives a very attractive concept for future dense city deployments. It allows for great capacity, dynamic management of capacity, and avoids the need for extensive fiber deployments and significant infra-structure projects. Instead the infra-structure is flexibly deployed and configured, and no longer is a limiting issue on the location and deployment of cells sites.
Future Testing for 5G
As the network concepts and technologies develop for 5G, so the corresponding test methods and processes will evolve to match this. Future 5G test methods will need to provide a high confidence to operators that the technology and services are implemented according to specification, and that the quality of service is matching to the requirements of the application or service being delivered.
A fully data-centric 5G network with a very wide and diverse set of applications to test would require a massive effort in standalone testing. Test automation, monitoring and built in test systems will be essential for analyzing properly the performance of such a network. In addition, the emergent solution to use Ultra Dense Networks (UDN) for interconnecting the radio access elements with the backhaul architecture using cloud networks will enable the development of cloud based test services for testing everything from everywhere. So, although 5G will introduce many new test requirements and challenges by the use of SDN/NFV and cloud services, this same technology can also be used for creating new test solutions that address these needs. With this in mind, cloud solutions are seen as both the new demand and the new solution for 5G network testing.
Test and Measurement challenges
One aspect that is expected to characterize 5G networks is that it is not a single technology or feature, but that it is an integrated set of technologies and features design to work together in an optimum way to support a wide range of applications and use cases. From a testing point of view, this means that as well as testing of the new technologies being implemented, there will be more of a need to test the many different combinations of technologies and network elements to ensure correct interoperability. This is further complicated by “virtualization” and the fact that specific network functions (the normal way in which network elements are tested) may no longer be physically realized in a specific piece of hardware, but may be moved around the network or split across different parts of the network. This virtualization of the network and its functions will require a corresponding virtualization of the testing methods and tools to be used.
Key parameters for testing 5G
- Traffic density Tbps/km2.
- Connection Density. Connected users and active users.Connections/km2.
- User data throughput at application level, both peak and average.
- Latency end to end through network, and round trip time (RTT).
- Reliability (error rates)
- Availability (coverage, handovers, mobility, capacity).
- Battery Life (mobile devices) and power consumption/power saving (network elements).
- Waveform/access spectral efficiency.
- Resource and signaling efficiency and overhead ratio.
The test industry already has available basic RF measurement tools (spectrum analyzers etc) that support the design and test of radio units at these frequencies, and they are already in use in the industry. But, cellular network testing introduces needs for more complex network simulators, and these have not yet been developed for millimeter wave bands. One of the key challenges in this area is that it is expected that a 5G network will use a wide range of operating frequency bands, and may dynamically switch between them, but most user devices are expected to be optimized for specific use case scenarios. Therefore the devices will not have a standardized or complete set of radio interface technologies, instead each device will support a specific subset of technologies. This is similar to the 4G situation of devices having specific radio band support rather than support for all 40+ standardized frequency bands, but now with an added second dimension of millimeter wave bands and technologies to support. Network simulators have supported this issue in the below 6 GHz band by using a flexible open RF architecture, but this may not scale up to millimeter wave bands and to different access technologies.
Along with the need for more frequency bands to support, there is also a trend towards wider bandwidth channels to support higher data rates. Whilst it can be expected that the core technology for such wide bandwidths will naturally be developed for the network elements (e.g. base station transceiver technology), there are specific issues for network simulator technology in terms of having a generic implementation that scales for different networks, rather than a specific implementation that is needed for real network elements. This requirement for additional flexibility in configuration becomes a key challenge in designing the RF performance and connectivity in a network simulator, being able to support the wide range of multi-carrier signals expected from a 5G network.
A second challenge related to this area is the issue of “connector-less devices”, as large arrays required for massive MIMO and designs for millimeter wave will be constructed into compact form to be cost/deployment efficient. This compact form means that there is unlikely to be space for RF connectors or test ports to attach test equipment. So the expected scenario is for “connector less” testing, and hence the need for more “over the air” OTA testing. OTA is an established technique currently, but uses large and expensive RF chambers, and is a relatively slow process. The industry challenge is to introduce more compact and cost effective OTA test solutions that are scalable for both RF (below 6 GHz) and millimeter wave band (24-86 GHz) use. Massive MIMO also introduces further challenges in terms of compact and affordable test systems to measure beam forming, beam shape, composite gain, and other antenna parameters related to the synthesized electronic beams. Studies are currently under way to evaluate near field measurement techniques to enablea compact test system, whilst being able to extrapolate to the far field antenna performance.
One of the most challenging technologies discussed for 5G is the concept of full RF duplex, where the time/frequency separation between transmit and receive path is changed to simultaneous use. This offers a very attractive benefit in terms of available capacity, but introduces some key challenges for testing and simulating. For such testing we need to provide a very high isolation test environment that is capable to measure beyond the design specifications of the 5G system. The core technology required to implement the full RF duplex concept is expected to give enough performance for this. However, the challenge comes in that the test equipment must usually be flexible for use in different RF bands, whereas full RF duplex technology is expected to be optimized for specific frequency bands. Therefore the test equipment must meet a critical cost/ performance point in trying to implement a cost effective test solution, balancing the band specific RF needs versus the cost effectiveness of a general solution. Without such a cost effective test solution, there can be significant cost barrier to having a wide deployment of the technology in the industry.
Future Test Instruments
Network test
In a high level perspective, NFV and cloud will require more intelligence and analysis in the test instruments as the network becomes more dynamic or alive. The physical elements in the network will be divided into specific/dedicated hardware such as antennas and radio transceivers, and general purpose hardware such as server/ computing platforms. Each of these will have specific hardware related characteristics to be tested, such as radio emissions or sensitivity, or processor data rate or latency. The protocol aspects will be tested separately, initially in a virtual environment (i.e. inter-action of protocol stacks housed on software platforms rather than on target hardware). Finally, the integration of software/protocol onto the hardware platform needs to be tested. In SDN/NFV, this then becomes more challenging as the hardware platform will not be a specific platform optimized for the required protocol. This will mean that the performance of the network will need to be constantly verified as each virtualized network function is located or moved to different hardware elements. This type of monitoring is expected to be made using independent probes in the network, which are independent of the network hardware platforms and can quickly detect any performance issues.
As with current networks, it will be necessary to test and monitor the different data links, to identify any bottlenecks or restrictions in data capacity, so the overall capacity can be optimized and maintained at the required levels. But with SDN/ NFV these statistics need to be provided in real time so that any dynamic effects due to software configuration algorithms can be detected. This is because network functions (and hence any problems) may not statically lie on a specific hardware component or use specific network links but could be software redefined by the network, and so a real-time monitoring system will be required to identify when a configuration is used that does not meet the performance requirements.
RF component and module test
5G will bring devices operating in much higher frequency bands, wider bandwidths and multi-carrier designs. More and more integrated technology is expected, which will produce a more complex analysis for interference limited devices.
User device / terminal test
Testing a 5G full-capable handset will become a very extensive process if we extend current handset test procedures. We can expect many more band combinations or RF duplex scenarios, massive number of transmitters and increasing types of receivers, leading to many variations in devices that will constantly be increasing. User devices probably will no longer be tested using cables because ‘Over The Air’ (OTA) performance will be essential in the testing process, for example in Massive MIMO testing. More complex shield boxes and chambers will be required, even tunable to very different frequency bands. Alongside this, the simulation of multipath using fading simulators will grow further in importance, and also in the technical demands and capability required in such equipment. As the MIMO and HetNet algorithms become more complex and powerful, then correspondingly powerful fading simulators are required to test and verify the algorithms. The integration of the network simulator and the fading simulator then becomes a critical issue, to enable accurate set-up of required scenarios.
Application testing within an end-to-end environment will increase in complexity and customer experience testing may become embedded into the networks and devices, for example using smart software agents. Finding the bottleneck in the throughput, in complex QoS and multi-link transmissions, becomes far more challenging. As the uplink and downlink, control and user planes, are each separated, then the limitationin data throughput can come from any of these areas. So even when the downlink user plane has enough capacity to deliver the required data rates, the downlink control plane needs corresponding capacity and latency to schedule it, and the uplink needs low enough latency to carry the fast HARQ feedback and MIMO reports (or similar control loop data) to maintain the high throughput.
