Testing 5G networks: challenges and opportunities
Until 4G, mobile phones were the primary devices connecting to the network. With 5G, this is going to change dramatically.
Unlike 4G, the impact of 5G is not limited to the telecom industry but towards a plethora of industry verticals. In the near future, every device present in our homes, almost all machines in our factories, cars, robots, and medical appliances will be capable of connecting to the network.
This blog post analyzes the impact of 5G on test tools, test strategy, test planning, testing methodology, integration and deployment both for vendors and CSPs.
Due to its broad range of use cases, mm wavelength support, extremely high throughput, ultra-low latency etc., testing 5G networks would be almost as complex as designing the network itself.
Certainly, one size fits all testing approach just won’t work for 5G.
5G’s complexity can be a blessing in disguise as it presents an opportunity for CSPs and vendors to evolve their traditional testing methodologies and come up with new testing and assurance techniques. Faster and more efficient testing methods will be the key for CSPs to differentiate from the competition. With 5G’s service based architecture and hundreds of use cases, validating a 5G network for a good user experience becomes really tough. Overhauling the test strategy is the key to ensuring fast time to market products thereby leading the 5G race for the CSPs and vendors.
Traditionally, with 3G and 4G the main focus was to test the new features, throughput requirements, radio conditions, handovers etc.
Testing was mainly done with test tools which simulate/emulate from few UE to few thousands of UE’s.
Vendors used to perform feature, stability, capacity, RF based test etc. before a product was delivered to CSP who then do drive tests, scenario test, radio antenna calibration testing before the products (e.g. RNC, eNodeb) were live in the network. Based on issues found during real network testing, bugs were fixed in next release of the product or if it’s too critical then in few days.
Why 5G needs a test strategy overhaul
Building a test strategy for 5G is a challenge in itself. Use case driven test architecture approach is one way we can simplify the test strategy. Since testing requirements for millions of Internet of Things(IoT) sensors, for autonomous vehicles, for ultra-high throughput clubbed with cloud RAN/CN, massive MIMO, beamforming etc. are all different.
CSPs, telecom software providers and vendors needs to be more agile and follow faster software development, delivery, testing, integration and deployment model as being followed by internet giants like Amazon or Netflix. It’s a great opportunity for some of the tier-1 CSPs to move from being a traditional telecom service provider to an Internet giant with agile and lean operations.
For faster 5G deliveries and bug free software here’s a proposed five-point strategy telecom companies should follow:
- Use case driven testing strategy: With hundreds of use cases and tens of industries already on the 5G radar, it’s important to identify the top use cases and industries. CSPs should model top 10-15 use case scenarios and make the test strategy around it. Top 5G use cases e.g. very high throughput, sub millisecond latency, massive IoT connections, remote surgery, enhance video services all have different test architecture requirements. In another report by Ericsson, clustering analysis resulted in 9 use case clusters, covering almost 90 percent of the addressable 5G business potential opportunity.
- Automation: Analytics driven automation is a prerequisite for 5G test environments and assurance. Network slicing is one important use case of 5G. Slices will be dynamically created, used and deleted. Also, slices can only be created for few seconds to few hours depending on the usage and requirements. Managing such dynamic behavior is possible only through automation. Since changes will happen faster than human can comprehend hence analytics and machine learning based automation is a key to 5G system testing and deployments.
Automation should be such that the test systems proactively sense when System Under Test(SUT) is going to go down or have major alarms or degradation of service and automatically suggest corrective actions.
- Continuous everything: For 5G, continuous everything is not just a technical cliché but probably the most important change which telecom vendors and operators need to adapt as they move into next decade. 5G’s unprecedented complexity needs a DevOps based approach with continuous testing, integration and deployment across the lifecycle. Static test strategies as had been used in the past will no longer work in this fast paced DevOps era.
Micro services based test libraries and APIs would be deployed allowing faster and flexible integration with new releases and product from different vendors. Continuous integration will be dynamic in nature with test suites and cases to be updated regularly.
- Cloud native software based Testing: 5G testing architecture must be software based with different test suites, test cases, configurations, test scalability etc. should be done with changes in software or configuration in static files associated with test suites. These applications should utilize services and underlying infrastructure provided by cloud. Efforts should be made to minimize the use of proprietary hardware thereby enabling on the fly test runs and faster test times. Test tools and methodology should support reusability of test components, protocols and services. Test tools protocols and services can be containerized to make them loosely coupled enabling faster development and deployment.
- Optimizing the legacy testing framework: Legacy testing framework using in 3G/4G networks can also be optimized to be part of the new 5G test framework thereby enabling legacy, inter RAT and Inter System Handover testing scenarios. Existing debugging and logging mechanism also need to be improved upon for faster resolution of bugs found during testing of System Under test(SUT). A unified logging mechanism across all test tools e.g. functional, load, performance, RF tools will greatly improve the end to end(E2E) debugging of the entire network simulation as well as for the real network testing.
5G deployments will pick up pace in coming years going into next decade and by 2023 about 19% of all data traffic will be over 5G. 5G will open up unlimited opportunities for new use cases, businesses, devices and applications. However, the journey from 4G to 5G is complex and is not just a move to a new generation of mobile communication but rather a paradigm shift which requires new ways of testing and validating the network. Both CSPs and vendors need to adapt to new testing methodology and assurance techniques for a faster 5G deployment and making this journey a success. It may be a challenge adapting to new methodology but it will provide CSPs a chance to transform themselves into new age digital service providers.