As of today, more than 100 commercial 5G networks have been launched worldwide. North-America, Europe and some countries in the Middle-East have already deployed 5G technologies. At the same time hundreds of operators globally are investing in the 5G, so deployment paste seems to be faster than the for the previous mobile technologies. The feedback from the first deployments is positive and speed tests provided fantastic results in terms of throughput and latency. Opensignal has conducted a study earlier this year with following results: 5G was able to provide 14-times higher download speeds than 4G. So 5G is not just a promise anymore, it is real!
As we serve customers who operate the most demanding networks and connectivity applications, we have accepted the challenge of making network performance measurements better for the new reality.
There are definitely many questions related to 5G transition and its impact on the service quality assurance. In this post we will answer 5 of them:
1. What will SLAs for business service look like in the future?
Surely, SLAs will contain some KPIs related to the network performance, which we already have today. But the main goal is to take all existing SLAs one step further since the quality of the experience is really critical. It is only logical that there will be additional quality-of-experience KPIs added to describe how the users of the service are really experiencing it. Instead of just providing monthly reports we would need to offer real-time status of network performance to the customer.
2. Is it possible to monitor 5G SLAs based on network slicing?
The answer to this question is definitely yes. The operator may deploy both the physical and virtual sensors in the portfolio for a network slice and he has all the advanced SLA monitoring capabilities of the PULScore system, such as flexible SLA templates, real-time SLA status, monthly SLA reports and customer SLA portal available for network slicing as well.
In addition, our mobile user experience features can be used to further provide deeper insights into the end user experience beyond the traditional network quality parameters for a network slice.
3. Can the data Creanord is measuring and machine learning combine to predict network outages?
To start with, we have to note that “slow is the new down” in the connected world we are living in today. Traditionally, operators have been very good to handle short-term outages in the traditional sense when, for instance, a data link is cut. Networks would react automatically, redirecting the traffic along a new route. In the current connected world, we need to consider also delay and jitter components to determine whether the service is working or not. Too big delay values for a time critical service such as Voice over IP calls means that the service is not working even if data is still flowing. It is in these scenarios that PULScore excels, network devices are not able to handle these types of scenarios.
Machine learning and artificial intelligence in the PULScore system offers a lot of value in predicting network outages when considering the holistic view of the network performance (loss, delay and jitter). The PULScore system will warn the user if any of the KPIs are getting close to a level where the service is affected. This can be fine-tuned to allow for sufficient time to react to the issue so that the end-user is not affected.
One more advanced machine learning feature in PULScore is the capability to learn the normal traffic behavior of the network by using historical data. This knowledge can then be used to inform the user or an umbrella system if the traffic patterns deviate from the normal baseline behavior. This is very useful in predicting a network outage ahead of time as there may be indications in the data long before the service is actually degraded. Thus, the operator may react and make modifications to the network before the service is actually affected. Baselines are effective in predicting potential network outages hours to a few days before they are actually occurring.
Another way to predict outages is to look at trends within the network. The intelligent algorithms within the PULScore system can calculate trends for the service KPIs and from the trendlines, you see when a potential issue may arise in the networks. Trendlines work best on longer-term traffic patterns in the network and is effective in predicting network outages days to weeks before they are actually occurring. One final, but very important, use case is the scenario where there is already an AI/ML capable system in place analyzing data from multiple sources. For these cases, the PULScore system can act as a KAFKA producer to feed the AI/ML capable system with real time information from the network over the KAFKA bus in order to enrich the data being analyzed.
4. What is the key difference between Creanord and equipment providing management software like HUAWEI?
One of the differences when comparing Creanord to equipment providers is that our main focus is on providing a full suite of measurements and features for measuring network performance combined with advanced analytics and presentation capabilities. The suite of supported measurement types that include both basic network quality measurement as well as quality of experience measurement types, the micro-second level accuracy of the measurements as well as the advanced analytics and presentation capabilities powered by artificial intelligence and machine learning are essential for us in order to provide a complete network picture in terms of the performance of the network and the services.
In addition, interoperability is key to be able to provide a full network view. Our measurements are interoperable with all major network element vendors. Network element vendors, on the other hand, focus more on other aspects, so you are not getting the full suite of measurements or interoperability with other network element vendors. Scalability to cover a large part of the network with one probe is also one big difference to equipment providers.
In terms of scalability, the router vendors and equipment providers typically support a fairly low number of TWAMP sessions per router/box, some hundreds of sessions per device. When looking at some of the mobile backhauling 5G deployments you will want to measure connectivity from a central location to thousands of base stations. With this scenario, you can easily have tens of thousands of sessions per one location and it does not make commercial sense to stack the routers for measurement purposes only. Thus, in addition to several technical reasons like high accuracy and a full suite of measurement capabilities, it also makes commercial sense to have dedicated devices for a quality assurance.
5. Are there 5G networks where Creanord has deployed PULSure solution as of today?
There are already several 5G deployments in Scandinavia, Middle-East and Asia. And we are working on the next ones.
As it was already said there are many questions on 5G service assurance and transitioning to 5G. Here we have covered only 5 of them and we are happy to answer any other questions you might have. Join the conversation by contacting us or booking a call.