Ericsson: Automating traffic balancing across cells

Challenge: When individual base stations are overloaded by traffic, the ensuing congestion impacts the network performance and the end-user experience. Mobile operators need to be able to predict when base stations will become overloaded so they can proactively hand over traffic from the affected cell to neighbouring cells.

Solution: Ericsson has developed an AI-based solution to predict traffic-related key performance indicators (KPIs) for each cell for the next hour. Based on the temporal structure of these KPIs, the solution employs a configurable rule-based classification module to identify so-called victim cells. It then hands over traffic from these victim cells using A3/A5 offset mechanisms to neighbouring cells.  The solution is designed to be agnostic to cell and band addition/deletion. 

Impact: The solution is designed to boost overall traffic throughput, thereby improving the end-user experience. It also makes greater usage of the available network capacity and spectrum resources. 

Topic: Network optimisation, and predictive and analytic AI

Cost: Medium

MNO: Ericsson