LAUSR

Mobility management in 5G networks is a very challenging issue. It requires novel ideas and improved management so that signaling is kept minimized and far from congesting the network. Mobile networks have become massive generators of data and in the forthcoming years this data is expected to increase drastically. The use of intelligence and analytics based on big data is a good ally for operators to enhance operational efficiency and provide individualized services. This work proposes to exploit User Equipment (UE) patterns and hidden relationships from geo-spatial time series to minimize signaling due to idle mode mobility. We propose a holistic methodology to generate optimized Tracking Area Lists (TALs) in a per UE manner, considering its learned individual behavior. The $k$k-means algorithm is proposed to find the allocation of cells into tracking areas. This is used as a basis for the TALs optimization itself, which follows a combined multi-objective and single-objective approach depending on the UE behavior. The last stage identifies UE profiles and performs the allocation of the TAL by using a neural network. The goodness of each technique has been evaluated individually and jointly under very realistic conditions and different situations. Results demonstrate important signaling reductions and good sensitivity to changing conditions.