5G testing is, by no means, anything like what has been practiced so far. Because 5G has brought millions of things that are functioning at different speeds, streaming different data in addition to many more complex tasks being performed in parallel. Service providers and vendors have to be agile, faster and more efficient with their testing to be able to stay ahead of the curve. It also could be an opportunity for them to compete with the Internet Giants.
In this concluding post of our 5G CPE Test Scenarios series, we pick the final set of test case scenarios
For in-device 3GPP/Non-3GPP interworking algorithms to work, DPI algorithms should identify the traffic correctly. For example, a Zoom call will have live video and conversational voice components. If the DPI algorithm identifies only the video traffic, a bearer with QCI7 will get established. This bearer may not support conversational voice traffic which need QCI1 bearer. To test, its is important to run different types of traffic and check the output of DPI algorithm.
Voice Calls Under Heavy Load Condition
There is one traffic that the 5G world seems to take for granted – voice. While the bandwidth requirements of voice traffic is minimal, the real time characteristics of voice calls ensures that any slight disturbance gets noticed by the end user. Eventually, voice call quality is non-negotiable for the end user.
If there is no dedicated bearer for voice and the calls go through the existing best effort bearer such issues get noticed when traffic load get heavy. To test this, one can bring the 5G CPE load to near highest levels and start few voice calls (Skype for Business, VoIP, etc.) to measure the call quality.
Wi-Fi client mix
In the real world there is always a mix of Wi-Fi clients (11n, 11ac, 11ax). Currently the percentage of 11ax clients will be less. This percentage will increase over the next few years. The legacy client line 11n (Wi-Fi 5) is expected to be in use for many more years to come since Wi-Fi connected household devices like air conditioners, refrigerators, washing machines etc prefer 802.11n due to support of longer range 2.4 GHz channels and lower price points for Wi-Fi modules.
Due to this reason it is important to test end to end performance with different Wi-Fi clients mix. Presence of legacy clients in the network should not impact the performance of 11ax clients. (Note: We have seen instances where presence of 11n clients causes the 11ax throughput values drop drastically).
Run throughput, video, voice, and mix traffic tests with a mix of clients as shown below. The client mix is just a suggestion, it is recommended to get the client mix closest to what is seen in target markets over different time periods.
Performing quick benchmark tests across different CPE models and software releases re-evaluates the features of maximum interest. You can pick the CPE best suited for the market. For example: CPEs for school/college campuses might need to handle more video, multiplayer gaming traffics. CPEs for office use case will need better support for office productivity tools – Skype for business, FTP, emails, etc. CPEs from two vendors that uses the same 5G and Wi-Fi chipsets might behave very differently due to the custom in-device algorithms. For each CPE market (home, office, college campus, etc.) run most commonly used traffic at scale and provide appropriate weightage. This helps to provide a meaningful score for the CPEs from different vendors and help in faster decision making.
By measuring how external factors (weather in different cities, obstacles in the 5G path etc) affect the end user experience, a smooth end user experience becomes a reality. To elaborate, external channel conditions are different in each city where 5G is deployed. Factors like weather (rain, snow, etc.) and obstacles impact when mmWaves are used in backhaul. Field and deployment tests can ensure that the end user experience is maintained across the cities under different local weather conditions and other factors.