Anomaly detection in 5G beam propagation

Abstract: Advancements in today's technology has motivated invention of faster mobile communication systems. The fifth generation mobile network, 5G, is the latest version made by the third generation partnership project (3GPP) and expects to both increase connection speed and reduce latency, which eventually will make it applicable in supporting state-of-the art technologies such as virtual reality and self-driving vehicles among others. A major part of the 5G system is beam management, the system that controlls how signals or beams are assigned to user equipments such as phones and computers. Previous projects have looked at how machine learning could be used in order to improve beam management which has worked well. However, noise in the beams makes algorithms unreliable at times and signals unstable. The objective of this master's thesis was thus to apply machine learning in order to decrease the noise in the data, improving performance of beam management further. The machine learning algorithm that yielded the best results was a clustering algorithm called Density-based spatial clustering of applications with noise together with K nearest neighbour. The results implied that by applying the machine learning algorithms and hence avoiding noisy data, higher signal power, higher data transfer speed and lower control signalling overhead could be achieved.