The amount of data generated by autonomous driving must be analysed and examined for anomalies.
The amount of data generated by autonomous driving must be analysed and examined for anomalies.
( Source: gemeinfrei / Unsplash)

Data Services

5G: Turbo for Data Analytics

| Author/ Editor: Stefan Schwarz* / Florian Richert

The hype topic 5G not only drives autonomous driving and applications in the Smart Home and IoT. Data analytics will also experience a boost due to the availability of large amounts of data - thanks to 5G almost without latency.

The hype topic 5G not only drives autonomous driving and applications in the Smart Home and IoT. Data analytics will also experience a boost due to the availability of large amounts of data - thanks to 5G almost without latency.

The new 5G mobile communications standard is intended to make networks faster and more powerful - and thus open the door to new services. These include Internet of Things applications such as Smart Home, Augmented and Virtual Reality as well as vehicle-to-vehicle communication and robotics. In all these cases - and countless others - devices and IT systems must be able to communicate with each other in real-time. By enabling 5G to transmit large amounts of data with low latency, it has a direct impact on how and in which scenarios data analysis will be performed in the future and drive innovation: How can 5G's technological potential be leveraged through data analytics? And how can Data Analytics itself contribute to a secure and stable 5G network?

With data towards autonomous driving

The more reliably and quickly data is transferred, the more reliably and quickly it can be analyzed. 5G is therefore clearly an "enabler" for data analytics in the IoT area. Take the example of networked and autonomous driving: Safe traffic is only possible if vehicles over 5G can communicate with each other "peer-to-peer" within milliseconds and exchange and analyze data on driving behavior such as distance, speed, and braking.

Ensuring the availability and analysis of large amounts of data without latencies is essential for artificial intelligence, which enables a car to make decisions for us humans in real-time. For this reason, one of the most important tasks of Data Analytics is to enable the automatic reaction to anomalies in the data sets. Only if a system detects these anomalies - for instance, if the car in front of it brakes unexpectedly or a person falls onto the road - can it react accordingly and learn from them for similar scenarios in the future. Integrated cloud edge analytics concepts and crowdsourcing aspects play an important role here.

As 5G is at its beginning, artificial data is currently often used to detect anomalies due to a lack of real data. In test runs, these are artificially integrated into the data records to check the accuracy with which the systems can identify them. The car manufacturer Volvo is already analyzing real data from around 500,000 incidents per week of its vehicles with the agreement of its customers to predict errors and make networked driving safer.

Automated network maintenance with data

The car of the future runs on data. This requires a stable network. Telecommunications providers are already using data analytics to maintain their mobile networks. 5G can make this task more efficient and easier by using more powerful analysis tools and machine learning models. These tools make it possible to automate remote maintenance and operation, for example by helping to prevent undershooting and overshooting. How exactly? Radio waves emanating from radio towers are three-dimensional structures that change continuously over time. They ideally lie next to each other and overlap only slightly to guarantee good reception. Weather conditions such as rain and wind can harm the position of the individual radio towers and result in an under-shoot. In this case, the signal does not reach the other radio wave at all. A radio hole is created. In "overshooting", however, the radio waves overlap and influence each other. Reception interference occurs. So what to do? Today, modern transmitter systems can be automated remotely and set in real-time in such a way that both cases do not occur in the first place (keyword "self-organizing networks"). This form of predictive maintenance requires in-depth data analyses (Prescriptive Analytics), which can be used to predict disruptions and make preventive recommendations or even perform autonomous actions. By linking and examining important parameters such as weather, location, data volume and many more, realistic forecasts can be made to optimally align the transmitters in advance.

Using data to schedule the 5G Rollout

5G rollout can also be optimally planned using data. 5G is associated with very high costs for telecommunications providers. The auction for the 5G frequencies in Germany alone cost the four participating companies almost 6.6 billion euros. Also, there is the investment in the network expansion itself, so that overall costs can be assumed to be in the tens of billions. In this context, savings opportunities in the single-digit percentage range already represent significant sums. By using data analytics and artificial intelligence methods, telecommunications providers can obtain a much more accurate picture of the regions in which 5G is needed and where the greatest return on investment for 5G can be achieved. To determine the greatest need for 5G, you can use historical data on 4G usage and other technologies.

High expectations

Most mobile phone companies plan to roll out their 5G networks in Germany within the next one to two years. Strong competition is also accelerating the roll-out - and driving the transformation of an entire industry. Telecommunications providers have long been using Data Analytics not only to improve their existing services. They turn the data they transport from A to B into a service themselves - and position themselves as providers of data analytics.

This article was first published in German by Industry of Things.

* Dr. Stefan Schwarz, Director, Industry/Business Consulting and Solutions, Central Europe, UK & Ireland at Teradata.