SIEMENS

Complex systems such as weather, traffic, stock markets, and biochemical processes in the body don’t operate in accordance with chance but are instead subject to non-linear laws. The individual components of such systems mutually influence one another and are continually resorted. Dunes organize themselves, for example, as do clouds, ant colonies, the light pulses in lasers, and signals in the brain. The complexity of a system increases in line with the number of elements it contains, the number of connections between these elements, and the degree of non-linearity in the relationships between the connections. As a result, doubling the intensity of a signal doesn’t necessarily yield twice the effect, for example, but might instead lead to a four- or even eight-fold increase in signal strength.

The theory of complex dynamic systems nevertheless often allows trends affecting several systems to be modeled on the basis of just a few parameters. For example, an intelligent traffic guidance system doesn’t need to know the actual behavior of each driver on the road in order to be able to forecast certain traffic waves or gridlock. Instead, such a system is trained to discern a trend at the right moment from fluctuating traffic density patterns and then adjust traffic light sequences or tunnel entrances as needed.

The brain offers one of the most interesting examples of a complex system. When people learn new things, the nerve cells in their brains autonomously create new connections and form new neural networks. The stimulus patterns transmitted by sensory organs ensure that ever-more complex behavioral patterns are created as a result. The knowledge gained from brain research and studies of complex systems in general has also led to a paradigm shift in computer science. For quite some time, scientists thought that complex systems could be controlled only with superordinate programs — but today we know that many processes in power generation, manufacturing, traffic guidance systems, and logistics can be managed by neural networks that function in a manner similar to the way nerves link up in the brain. The big advantage here is that such artificial neural networks can learn from examples generated in real time and respond flexibly to changed conditions.

Neural networks are used today to control the operations of power plants. For example, the gas turbine that Siemens built in Irsching, Germany, for what is currently the world’s most efficient power plant is equipped with thousands of sensors that continually monitor air pressure, exhaust gas temperatures, and emissions. Software systems modeled on the human brain then evaluate the data and learn autonomously while doing so. The measurement data isn’t used to optimize just one power plant, however; using a sort of swarm intelligence, it’s also possible to link several plants that continually optimize themselves during operation on the basis of experience.

Similarly, in the future “Internet of Things,” many devices will be able to exchange data with one another, access Web services, and interact with people. Clothing will tell washing machines what temperature needs to be set, for example, and cars will communicate with one another in order to prevent traffic jams and avoid accidents. The power grid of tomorrow will link thousands of energy producers and consumers, software agents will purchase and sell energy autonomously, and components in factories will be fitted with smart labels that enable them to organize themselves and control production processes via radio communication.

Industrial facilities will also collect data on entire product lifecycles in order to optimize manufacturing, product operation and recycling, while integrated traffic and transport systems will combine various modes of transport in order to get travelers to their destinations as quickly and conveniently as possible. The images generated by state-of-the-art medical devices are already being interpreted by computers and then linked with information from knowledge databases in order to assist physicians with their diagnoses.

This Internet of Things will make it possible to access knowledge in a totally new way and also enable the development of new business models and services. A new Internet of Knowledge and an Internet of Services will be created in a similar manner. Most importantly, however, the number of objects and items linked to one another via the Web will increase rapidly. Market researchers at IDC, for example, expect that 15 billion online-enabled devices will be linked to one another worldwide as early as 2015, and this number will increase to more than 50 billion by 2020.