Deutsche Telekom now uses artificial intelligence in many different areas, ranging from customer service to the planning for expansion of the fiber-optic network, expansion planning to services for key accounts. Here are a few examples of relevant products and services.
Cyber defense with AI
The laptop has a life of its own, the smart TV takes an eternity when switching, some account debits are strange? Then maybe that's the work of cyber criminals. Cybercrime is becoming more and more professional and uses automated attack methods. That's why Deutsche Telekom uses AI in cyber defense.
Deutsche Telekom uses sensors built into its networks to analyze attacks. These register up to 50 million attacks daily. Overall, Deutsche Telekom's cyber defense analyzes about one billion security-related data every day for signs of cyber-attacks. Of course, even cyber experts were not able to manage this by hand. For this purpose, Telekom relies on fully automated help from AI. Computers are much better at detecting patterns and quickly identifying deviations that point to attacks. A valuable support for analysts in cyber defense. Telekom has developed and trained the AI itself, in cooperation with Telekom Innovation Laboratories in Israel. Telekom has been using AI for over two years now to protect the security of its own infrastructure and thus the security of its customers.
More information on the Telekom Youtube channel "Netzgeschichten".
Live view of the cyber attacks is available at www.sicherheitstacho.eu.
The CONNECT App – Making it easier than ever to always get the best Internet connection available
The CONNECT App frees users from having to juggle manually between Wi-Fi and mobile networks to try to find the best available Internet connection.
Drawing on the power of machine learning, the CONNECT app always "knows" how good the available connections are and when it's best to switch from a mobile network to a hotspot or vice-versa.
Users can choose between "Best connection" and "Prioritize Wi-Fi" profiles and thus maintain full control over their charges and connection speeds. Optional VPN encryption provides added security in public Wi-Fi networks. The app includes a range of additional useful features, including a speed-test function and a network map that shows nearby Deutsche Telekom hotspots and WIFI TO GO hotspots.
The app is available free of charge in the Google Play Store and the Apple App Store.
Tinka: A virtual service representative with T-Mobile in Austria
Tinka is a chatbot that may be likened to a search engine. On a monthly average, "she" chats with about 60,000 callers, answering some 120,000 questions in the process. To date, Tinka has learned over 1,500 answers, and "her" answers are continually being updated. Tinka is available to assist T-Mobile customers in Austria at any time of the day – right away, with no waiting time. Tinka appears on customers' screens as an icon depicting a young woman with long hair, and with a box below her for text input. She is able to handle about 80% of all questions put to her. When she can't answer a question, she forwards it to a human colleague.
Although Tinka is already in service, Deutsche Telekom developers are continuing to work with customers to adapt Tinka more closely to customer requirements. And they have continued to integrate new sources of information within her overall information pool. Examples of what Tinka is now able to do include supporting customers in setting up LTE-based home Wi-Fi networks and explaining to customers how to insert SIM cards into their phones. One of her special strengths is that she can forward customers with difficult queries to other, available service channels – i.e. to customer service agents. If no agents happen to be available, she recommends email as an alternative channel.
Tinka will soon be able to recognize customers and greet them by name. She will remember previous conversations and be able to refer to them. Tinka will also be able to know – subject to customer consent – what a customer has already viewed on the Austrian T-Mobile website. This will make interactions with her much more specific and more focused on customers' individual concerns.
Since fall 2016, Deutsche Telekom's operations in Germany have been using a "digital assistant" to answer caller questions on selected topics. At present, Deutsche Telekom customers interact about 50,000 times per month with "him" – regarding such matters as SIM cards, smartphones or Wi-Fi networks. As of 2018, "he" is being upgraded with an artificial intelligence system based on the IBM "Watson" system. "He" now learns on the job, automatically, and thus is getting better and better at handling customer inquiries and concerns.
Anyone who wishes to see what the new system is like is welcome to try it out here.
Chatbots are still far from being able to replace real human service agents in addressing complex concerns, however, because they are incapable of empathy, a key ingredient in any top-quality service.
"Hallo Magenta" – the Smart Speaker
Deutsche Telekom plans to begin offering its customers the Smart Speaker, an intelligent personal digital assistant that answers to "Hello Magenta."
The constantly learning Smart Speaker manages connected devices in users' homes, guided by voice commands. For example, it supports voice control of Deutsche Telekom's EntertainTV service, for actions such as changing channels and adjusting volume.
Simple voice commands suffice for control of Magenta SmartHome applications with the system, for actions such as dimming lights or adjusting room temperature. It also plays voicemail messages on voice command. To do so, it connects directly with the user's router, so no additional devices are required.
Our aim with the Smart Speaker is for customers to be able to speak normally to it, and soon forget they are talking to a machine. In other words, we plan to keep improving the Smart Speaker user experience.
The functionalities for which the system uses artificial intelligence include recognizing the wake-up word ("Hello Magenta"), understanding speech and recognizing context for spoken words.
The Smart Speaker provides the kind of voice user interface that users now expect from voice-controlled systems: functional – not obtrusive or "strong-willed."
T-Systems supports preventive maintenance
Deutsche Telekom is now cooperating with the German firm BS2 Sicherheitssysteme in the area of bridge maintenance. The move makes sense because BS2 Sicherheitssysteme has developed a digital early warning system for bridges, tunnels, buildings, and other infrastructure objects, on the basis of Deutsche Telekom's machine-to-machine network (NB-IoT). For each structure being monitored, the system employs a range of sensors that detect warning signals long before any problems become visible, by monitoring critical factors such as temperature, humidity and corrosion.
The new Internet of Things (IoT) system makes structures "smarter," safer and more durable. It helps minimize damages – and the resulting repairs and costs.
Power tools such as electric saws, drills and jackhammers also always need to be kept in good condition. In cooperation with the firm Toolsense and with T-Mobile in Austria, the company is developing smart IoT solutions, for tool manufacturers, that help enhance important performance criteria such as durability, wear-resistance and energy consumption.
The MyShake app turns smartphones into an early warning system
Researchers from UC Berkeley are working together with Telekom Innovation Laboratories on a smartphone-based earthquake early warning system. In February 2017, the team presented the app MyShake, which can pool Android smartphones to create a network of seismic sensors that can warn of earthquakes seconds before they hit.
The system works with the accelerometers that smartphones use in running games and other applications. The researchers have developed an algorithm that uses such accelerometers – which are three-axis acceleration sensors – to record tremors. When its sensor data fit the vibrational profile for an earthquake, the app sends the time, location and measured strength of the tremor to the UC Berkeley Seismological Laboratory for further analysis.
In addition, cloud-based software checks the incoming flow of data. The program confirms that an earthquake is happening if at least four phones detect shaking, and if the number of detecting phones exceeds 60 percent of all smartphones within a 10-kilometer radius of the presumed epicenter. Researchers then compare the data with those from conventional seismometers.
Smartphone owners will be able to actually receive warnings via the system as soon as enough people begin using the app and the system is operating reliably. In the current phase of the effort, therefore, the primary aim is to generate interest in the system's crowdsourcing approach and to get as many people as possible to use the MyShake app. The denser the network, the faster the system's earthquake detection will be. Currently, the scientists are hoping to launch an app update early next spring – following a test phase of about one year – that will then be able to send warnings to users.
Real-time forecasts of train arrival and departure times
T-Systems will soon be forecasting train arrival and departure times for Deutsche Bahn trains, including long-distance, regional and suburban trains. A learning algorithm will provide real-time forecasts for the more than two million stops made daily throughout the railway network, along with updated forecasts of available connections.
Via a smartphone app, as well as via monitors at rail stations, Deutsche Bahn customers will be provided with real-time train-schedule updates up to 90 minutes in advance.
During its calculation processes, and operating in T-Systems' data centers, the system analyzes geo-positioning data for all trains in service. Calculation runs, which take only seconds, produce real-time forecasts of trains' expected progress along their scheduled routes. The system's calculation algorithm uses a range of resources, including machine learning (artificial intelligence) processes. In training runs carried out nightly, at 24-hour intervals, the system's network model "studies" historical data. It is thus able to continuously improve its forecasting accuracy and ability to adjust to momentary rail-traffic conditions.
The system, which is based on a solution developed in-house by T-Systems and its subsidiary T-Systems Multimedia Solutions, is being upgraded and implemented via a project in cooperation with Deutsche Bahn.
Planning for expansion of the fiber-optic network
Under earlier planning regimes, each phase of actual network expansion had to be preceded by a range of different planning steps. Soon, such planning will be able to proceed significantly more rapidly and with shorter preparatory periods. Planning staff's procedures for checking and confirming planning will be greatly accelerated.
Planning itself will be carried out with the help of a special vehicle that, via various sensors and laser-scanning technology, gathers precise data about the environments selected for expansions. The so-collected data will be translated into georeferenced 3D image data. The system learns to recognize landscape features, such as houses, grass, trees, etc., in terms of their planning relevance, and uses this capability to enhance the network planning process. Needless to say, the system is also able to incorporate available reference data such as street maps. As a result, the system can rapidly produce precise proposals for ideal routes for subterranean cables. The project is being carried out in cooperation with Fraunhofer.
