As defined by the NGMN Alliance, 5G is not really another ‘G’, but rather an ecosystem of technologies, services and verticals required to intertwine and smoothly coexist with older wireless generations, as well as other access technologies, such as WiFi, toward a seamless user experience, regardless if humans using smart phones or if machines.
5G networks are defined to support three main use cases: extreme mobile broadband (eMBB), massive machine communications (mMTC) and ultra-reliable low latency communications (URLLC), each with different peak bit rates, latency, mobility and reliability requirements, as well as traffic load, connectivity density, battery life, and devices’ costs. Consequently, unlike their predecessors, 5G networks unleash an economic disruption of the wireless industry by enabling the evolution of a variety of verticals from extreme broadband wireless consumer and entertainment to smart cities, asset tracking, e-health (e.g. remote patient monitoring) to Industry 4.0, e-health (e.g. remote surgery), AR/VR and gaming.
For a 5G ecosystem to exist, significant transformations are required for networks and devices, as well as for the services delivered.
Based on the start of our detailed white paper, this blog post discusses the five most important considerations for new radio drive testing in this era of new technical challenges.
1. Different types of QoE
We must consider different types of quality of experience (QoE) from the ones we saw with previous ‘G’s.
First, QoE needs to be redefined for various Internet of Things (IoT) verticals and use cases. That is, QoE within the context of machines and different types of applications, from massive connectivity to ultra-low latency. Therefore, QoE will need different metrics to define service-level agreements (SLAs) as well as the machines’ ‘experience’.
Second, regardless of the 5G use case, QoE shifts towards a context-aware definition.
A context-aware QoE of service/application delivery to consumers and/or machines is based on a continuous learning and adaptation between the network and the device, based on network knowledge on the user/device (e.g preferences, profile, etc), as well as the device/user knowledge of the network behavior/status (e.g. load, traffic type).
All these involve artificial intelligence (AI) and/or machine learning (ML) based solutions embedded in the network and devices, as well as used for network orchestration, operation and performance management.
Since the AI/ML solutions are not standardized, the role of network testing increases significantly. However, there are two major caveats: it’s user-centric (QoE-centric: humans and/or machines); and it’s highly automated/autonomous.
2. Multi-mode programmable devices
Devices need to be programable and configurable, multi-band and multi-mode, all of which is achievable through advanced signal processing, AI/ML-based firmware, and complex frequency-sensitive frontend hardware.
These come with significant challenges on device performance (e.g. interference, overheating, antenna positioning), which can be tested and evaluated through user-centric testing (e.g. drive/walk). Thus, the devices’ performances can be visualized and analyzed in correlation with bandwidth and network configurations, while running various real-life services and applications consumed by users and machines, in different network and radio conditions, outdoor and/or indoor.
3. Variation of large range of spectrums and bandwidths
Let’s look at new and large variability spectrum (below 1GHz and sub-6GHz to tens of GHz), and bandwidth availability and needs (e.g. 40-100MHz and beyond 100MHz), which come with diverse coverage and interference challenges depending on the spectrum type.
Depending on the requirements of each 5G use case, various spectrums and bandwidths are allocated (Figure 2). Generally, with both licensed and unlicensed bands, sub-5GHz spectrum will support mMTC (e.g. U-health/wearables and smart home/city); 5GHz up to about 20GHz GHz URLCC (e.g. IIOT and smart vehicle), and eMBB expands its deployments above 20GHz spectrum.
Each of these scenarios are using various enabling technologies (e.g. carrier aggregation [CA], Bandwidth Parts, enhanced Licensed Assisted Access [eLAA] and Dynamic Spectrum Sharing [DSS]), whose overall performance evaluation and troubleshooting requires testing at device-level in order to optimize their benefits from the user perspective.
In addition, spectrum is divided for different services and levels of quality of service (QoS), which increases the number of tests needed towards a deployed site. Consequently, automated and remote device-based data collection and analysis become crucial during rollout.
4. Network flexibility and scalability
The newly introduced technologies (e.g. flexible numerology and slot structure) embedded in lower layers enable the flexibility and scalability of 5G network designed to support various uses cases with different requirements. In addition, virtualization of the RAN along with the core, as well as the utilization of Mobile Edge Computing (MEC) implemented at RAN layers, are the main enablers of URLLC-type services and applications.
However, the ultimate benefit of these new network characteristics on the user (or machine) experience of a specific service/application can be evaluated only using device-based measurements.
5. Network slicing, one of the most significant consideration for 5G NR testing
We move on to one of the most significant areas of 5G NR testing: service-based architecture enabled by the Control User Plane (CUP) split and network slicing to support eMBB, along with simultaneous IoT verticals (mMTC, URLLC categories) with diverse and in some cases opposite performance requirements.
The network slicing concept per se emerged earlier in 3GPP with the QCI allocation to various LTE bearers. With 5G the concept evolves into dynamic network slicing functionality, which enables low-cost real-time service configuration with the ability to deliver specific SLAs for specific groups of application or type of client, etc, thus opening the possibility of a vast array of new and potentially disruptive use cases, and vertical and horizontal opportunities across a wide variety of industries and sectors, such as C-V2X applications and autonomous drones.
The testing that’s needed is service- and user-centric, and is expected to involve two types of testing.
First, it is required to evaluate/benchmark and troubleshoot if a dedicated slice meets the SLA performance metric (KPIs such as availability, connectivity, reliability, latency, and throughput), as well as the slice’s efficient allocation (e.g. idle time or traffic load). This is likely to be achieved through various probes in the network and/or PM-type of data sources.
Second, within each slice, and/or combination of slices, the performance of a particular vertical (e.g. VR within URLCC and eMBB slices or V2X within URLCC) is met at device-level. In this case, device-centric testing enables evaluation and troubleshooting/optimization based on service/application requirements such as coverage availability, connectivity, reliability and latency in a connected car scenario.
A high-level summary of KPIs requirements per slice / 5G use cases is presented in a table in the white paper. In targeting these KPIs, the MNOs have to figure out what is cost-effective for them and then how to make those KPIs into policies that are implemented in their network. For example, the parameters and metrics used to tune a RAN to work well in Downtown City 1 may not work in Downtown City 2, due weather, for example; mmWave may propagate differently in each of those different cities.
As we can see, 5G-related disruptions have a strong impact on testing in general, with an important requirement for device-centric testing. Therefore, the ‘drive testing’ concept not only remains in place for 5G NR testing, but plays an even more important role with 5G NR due to the user/machine-centric design of 5G. That said, we should underline the fact that significant transformation towards automated and autonomous drive testing, as well as evolution to intelligent measurement clients embedded on devices, will play a crucial role in 5G NR testing.