How does machine learning help 5G networks?

Machine Learning

Machine learning is a field of engineering that has matured significantly over the past decade, thanks to the increasing power of computing systems and the availability of data. Unlike traditional systems, machine learning provides engineers with a tool that can not only be taught to recognize patterns, but also learn from its environment, which helps improve its performance over time.

In the early development of machine learning, it was mainly used for image and speech recognition, but this is changing in recent years. Machine learning is now widely used in fields such as medical diagnosis, stock market decision-making and even environmental control.

Channel Search

Wireless technology is incredibly complex, with each technological iteration adding an extra layer of complexity. The first wireless technologies based on radio signals would use spark gaps to receive the signal, while the next generation of radios would use diodes to demodulate the signal to extract the audio information. After several iterations of wireless technology, complex digital circuits combined with cryptographic functions would be deployed to keep information private.

Now that many devices are moving towards mobile technology, there is a high demand on cell towers with potentially thousands of simultaneous connection requests. To help manage this load, radio systems are deployed with channels that handle a very high number of devices per channel, and devices in one channel cannot interfere with devices in another channel.

However, finding a channel with low traffic can take a while, and using a good channel is often a factor of nearby devices and the environment. Since trial and error is used to select a channel, inefficiencies can lead to increased energy consumption and increased execution time.

Machine Learning Applications

To address this problem, a team of researchers at the National Institute of Standards and Technology (NIST) developed a mathematical formula that behaves like a machine learning algorithm.

Essentially, the formula selects a wireless network channel based on a priori experience rather than using trial and error. Since the system had a selected configuration in the past in relation to external factors, it can be argued that the same setup offers a better chance of functioning. The need for such a system stems from the fact that mobile networks are deploying a solution called License Assisted Access, which uses both licensed and unlicensed bands. This means that environments where both Wi-Fi and cellular devices are in use end up competing on channels, resulting in slower channel finding.

So if both antennas (Wi-Fi and mobile) use a machine learning-like formula to find good channels, they can operate independently to find the best solution. The formula, which maps environmental conditions, such as the number of transmitters and channels present, can reduce the number of attempts from 45,000 to 10, making it 5,000 times faster, according to computer simulations.

Machine learning’s ability to adapt to its environment allows it to improve performance over time. Such algorithms don’t have to be limited to audio and video applications. In theory, they can improve any process. Therefore, engineers should look at their own designs and try to identify situations that involve trial and error, and then see if they can be replaced with a learning algorithm.