What Businesses Should Measure When Evaluating a 5G Network
For businesses considering a move to 5G, the obvious benefit is faster speeds. However, speed alone does not dictate the value of the network to a business. Organisations are increasingly relying on their network for cloud apps, connected devices, real-time communications, automation, and heavy data usage, and as such need to consider a broader set of factors when assessing the network’s value to the business.
The broader 5G Technology Market is evolving in parallel with a rising interest in private networks, industrial IoT, edge computing, and increasingly connected workplaces. These trends are influencing how organisations are approaching connectivity. They are looking at the network not only in terms of how they can use it to access the internet but also in relation to how it can support the applications, devices and processes that lie at the heart of their daily operations.
Begin with the business requirements.
The first phase of evaluating a 5G network should not start with speed tests. Instead, it should begin with understanding what your organisation needs the network to do for it.
A connected scanner running within a warehouse has distinct requirements to automation equipment within a manufacturing setting, and a retail environment using smart devices managed by remote workers may view a connected healthcare facility supporting critical equipment differently. As a result, the network can provide dramatically different outputs dependent on deployment and utilization, despite utilizing the same network technology.
As a business, consider those applications running over the network, the number and types of connected devices, the physical spaces these devices may be located in, and the impact if connectivity falters. Consider real-time requirements, volumes of data to be exchanged, and whether it requires to be handled locally or in the cloud; these are all factors the 5G implementation needs to consider.
Such a use-case focused test regime ultimately provides a much more informative network assessment and takes the organization away from general “is it fast enough” tests into specific tests that tell it whether the network is fit for purpose in its specific circumstances.
Look beyond headline speeds.
Throughput, one of the most common network performance tests to be discussed, refers to the amount of data capable of being transported across a network in a given time span.
While throughput tests can be useful, these maximum advertised figures do not always give the full picture of an organisation’s connectivity. While a business may test a network when a single connection reaches a high speed and all variables are in its favour, reality may not always be as favourable when several devices are connected simultaneously.
Businesses must therefore look at a consistent, sustained throughput, not just the quickest one tested. These tests should be performed in multiple locations around an organization, and during different loads of use. Both upload and download capabilities need to be assessed.
Upload speeds might be increasingly significant given the rise of certain applications, like online surveillance systems, industrial controls, or video analytics, that may generate significant amounts of data that can be sent back up the line. By testing only download speeds, an organization might completely miss a capability required of their network.
Finally, network speed tests need to be done during busy periods. Should speeds drop off sharply as network load increases, the rapid performance from off-peak tests may mislead an organization into believing their network performs more admirably than in reality.
Understand what latency means for business applications
Latency is measured by looking at the time it takes a piece of data to go from one place to another. In basic use such as downloading or looking at a web page, small differences in latency will go relatively unnoticed; however, in some kinds of applications, both automated and real-time applications, Latency becomes far more important.
Some forms of technology may rely on speedy communication: manufacturing equipment, remote control systems, and collaboratively edited work, among other examples. In cases such as these, what seems like a minor delay to the casual internet user could alter the way the application responds.
Businesses, or their providers, look for average Latency in addition to Latency variations. Frequent bursts of Latency could make an application unpredictable. In these cases, the Latency would be higher than if it remained average.
The placement of the application also affects its Latency. Sending data to a large cloud service miles away may have more Latency. Edge computing provides a method of placing resources in locations physically closer to where the data is generated.
Research into industrial settings has tested the use of both 5G and Edge computing to work towards solutions to both automatic guided vehicles and advanced extended reality applications. (Nokia Bell Labs)
Reliability can matter more than speed.
For many organisations, a constant connection will carry more weight than a blazing fast one.
Imagine production where connected machinery requires a constant link in communications and the slightest blip can disable the entire operation regardless of network speed. Or logistics where constant drops by mobile equipment as it navigates a huge facility make it impossible to complete necessary tasks.
One way of assessing how reliable it is is network availability, but organisations need to look beyond a percentage of operating time for their services. Data like the duration of downtime can tell them as much, and more.
Testing should also find out about drop frequency, duration, and whether it occurs only in specific areas or conditions. Differentiating a network outage from an application outage will also be crucial, because a functional network doesn’t imply a functional cloud service, database, or application.
Test coverage in the real working environment
Although coverage maps are informative, they don’t tell an enterprise how the network will work in a specific building or at an industrial site. Radio performance can be affected by the building materials, machines, walls, stored products, physical impediments, etc. Large warehouses, for example, may not function like an open office space although they might be both located in the same geographic area.
Enterprises, for that matter, should always perform coverage tests where the actual equipment, employees, and devices are being used.
Production floors, stores, loading bays, outdoor yards, underground areas, transportation routes, etc. can be examples of these environments.
Coverage quality needs to be evaluated at the same time as application performance, since a given device will report a radio connection even when it is not performing to the capacity needed for the device it is using. Moreover, movement should not be neglected, as performance can change from static conditions when a device is moving from one location to another.
Evaluate how well the network handles mobility.
Mobility needs to be taken into account by organisations relying upon mobile devices such as connected vehicles, robots, mobile machinery, or anything that covers a defined site (such as a factory floor).
When a device moves, the network will need to be able to do so to swap it between one coverage area or cell and another. This switchover is known as a handover and an ideally undetectable one at that, so that the application does not suffer from disruption.
Organisations must analyzenetwork behavior at varying realistic speeds and routes for their devices and note areas where coverage cells overlap, as it is here where network limitations are exposed when static, but not visible when moving equipment is being tested with the network. Continuous live video feed, control, or even machines with full automation will become irrelevant, and with short gaps in live feed now a problem, mobile testing gives an accurate picture of how reliable a network is.
Measure packet loss and network consistency
Other factors besides latency can contribute to network response time. Packet loss and jitter can both impact an application’s performance.
Packet loss is when a data packet does not arrive at its destination. Certain applications can handle a few occurrences of loss; others will be slow and potentially unusable. Jitter is variability in the timing of packet delivery and can impact applications which requires consistent stream of data delivery.
These measurements are most relevant to voice and video and some time-critical applications. An application could have very low average latency on a network and a relatively poor performance with lots of variation.
In research on private wireless networks, we also identified that predictable behavior and constancy of networks in industrial applications can be as important as just speed of data (Nokia).
Consider the number of connected devices.
It’s also essential when looking at the performance of a network to know how many connected devices there are.
In one instance, what may happen with just some devices running at a given moment cannot accurately translate what may occur if hundreds of sensors, cameras, tablets, machines, and other endpoints are running.
Companies should determine current and future device needs. Companies should think not just about the numbers and volume of devices; they should also determine the data throughput characteristics for the devices.
This can mean an example like perhaps a building having a thousand sensors where there’s very little data associated with them, along with two or three video cameras each generating hundreds or thousands of times more data streams.
Test the applications, not just the network.
Perhaps the most relevant way to test a network’s performance is by looking at real applications.
A warehouse might be measuring how fast its inventory list is updated once a scanner sends off the update. A factory may be measuring whether work order information gets to the machines in a certain amount of time. A business in a call center could check for any disruption to calls when more and more users hop onto the network.
Testing at the application level should highlight situations that other measurements can’t. A database could be sluggish, an application may have server-side issues, or be poorly configured a situation that the network cannot overcome. Thus, a technically amazing network test doesn’t always mean you will see brilliant business results. The key indicator should be whether a business’s operational systems and employee-dependent applications are operating smoothly.
Treat security as part of network performance.e
Security has to be included from the very beginning of the evaluation rather than as an afterthought once the network selection process is completed. Organizations should look for ways in which devices will be authenticated by the network, how information is secured, and how access is controlled. They should also consider the extent to which the connected devices will be monitored and how security events would be detected and managed.
Network segmentation may also be required where different types of traffic share the same underlying network infrastructure; an example would be where operational technology and business IT need to be logically segregated or even individual types of connected devices.
Whilst encryption and authentication provide security to cellular technologies, overall security of a cellular deployment is architecture, configuration and integration with existing business IT-even a secure cellular network still requires effective endpoint protection, identity management, monitoring and governance.
Consider traffic priorities and quality of service
It is not the case that all business needs will require these features; these are only the features applicable, by case. Routine web traffic and video feeds, operating systems, and time-sensitive traffic may be all routed over the same network, and when volumes increase, the enterprise may desire some traffic to be given priority over others.
These measures, known as quality of service measures, provide a method to control application traffic in accordance with the type of application, and depending on deployment, may be incorporated into a variety of 5G architectures through network-slicing- whereby service requirements may lead to logically segregated services with different characteristics. (GSMA). These features must be seen as providing a viable business solution rather than simply being technically feasible for network service engineers.
Test what happens when things go wrong.
The only assessment that counts can’t be under ideal circumstances. A business has to know what will happen when network conditions change dramatically, such as a spontaneous increase in traffic, failure of some equipment, or a complete loss of connection to another portion of the infrastructure. Testing your infrastructure in this manner will quickly expose some obvious areas that aren’t in evidence under normal operating conditions.
For instance, you might need to see what happens when a device fails, or a hundred devices connect to their network simultaneously, for example, or you may need to know how long it takes for the network services to resume functioning.
A business needs to demonstrate robustness under certain circumstances, especially if the network is relied upon by operational technology. The impacts of potential network failure need to be understood if an interruption can actually impact production, logistics or critical services.
Include energy and operating costs in the assessment
In addition to technical performance, other aspects should influence your long-term view of networks.
When networks become widespread, so do connected devices. Deployments could result in significant amounts of connected equipment along with extra infrastructure elements. Companies must analyze energy consumption for network equipment and connected endpoints, especially devices running nonstop or on battery power.
Another key issue is operating costs. Installation, maintenance, monitoring, and the replacement of equipment all require money. So do networks themselves – software, security measures, and staff training all represent a cost of owning and operating any network.
Total cost of ownership offers a realistic perspective beyond initial deployment figures. Does increased connectivity impact key results that you could measure? Consider changes to equipment usage, equipment downtime, or process efficiency.
Build a measurement framework around real use cases
No one number tells a business if 5G is right for it. A more reliable assessment looks at several measurements and connects them to operations: throughput reveals if data can be transferred fast enough, latency indicates whether applications communicate quickly, and packet loss and jitter are about signal quality, while availability and recovery relate to reliability. Coverage and mobility testing look at network behavior across the physical environment.
Capacity testing helps a business decide if performance holds as user device numbers grow, while security testing checks if the network integrates securely into an organisation’s IT structure.
By testing the application, all of these technically complex measurements are linked back to the customer experience.
Why ongoing measurement matters
Evaluating a network does not necessarily stop once it has been deployed. Devices will grow, network traffic will be added and changed by software updates, and demands from business applications will evolve. Today’s fully satisfactory network may well suffer a shortage of capacity at a later date.
Monitoring the network enables organisations to pinpoint a potential deterioration before it becomes a business-affecting issue.
Baseline performance may be identified from the first assessment, with performance thereafter measured against this value. Network assessment becomes not a one-off technical exercise, but a continuous process.
What the future of network evaluation may look like
With connected systems becoming more advanced, businesses are going to be more focused on guaranteed application-specific network performance.
Edge computing, industrial IoT, automation, and private wireless are all becoming integrated concepts. This is due to the ability to bring devices and the computer power they need closer to each other. Industry has been researching how private wireless networks and edges are suitable for industrial use and a connected environment (GSMA).
This, among other things, is making basic questions like ‘what download speed will I get ‘ far too limiting to ask. Instead, the business must ask itself: do applications keep working when it is busy, are devices able to maintain connectivity while on the move, can the network recover if it goes down, and does the network operate reliably as the amount of connected devices increases.
Therefore, the “ideal measurement” framework will be tied to anorganization’ss operations.
Conclusion
More than download speed, a network’s performance must be gauged through throughput, latency, reliability, coverage, mobility, packet loss, capacity, security, and application performance.
All the right performance testing must start at the business usage requirement. After which, an organisation can specify appropriate targets against which it can perform relevant tests in a real-world simulation.
The assessment of a network should also include scalability over time, network resilience and the ongoing cost of ownership and any changes in future application requirements. Test for what actually matters to the business, and not headline test parameters, and organizations will be able to see a far clearer and more practical picture as to the fitness for purpose of the network.
About the Author
Roshan Kumar is a technology writer and researcher specialising in digital infrastructure, enterprise connectivity and emerging technologies. His work covers industry trends, practical applications and the evolving 5G Technology Market, helping readers understand how advances in connectivity are shaping modern businesses.
Read About : ted danson wife