PCI planning with graph neural network

In an LTE network, physical cell identity, PCI, is used to identify the serving cell by the user equipment, UE. The PCI allocation in an LTE network should be performed appropriately. Otherwise, it can result in poor network performance.One of the methods to allocate PCIs in an LTE network is graph coloring. In this method, each cell is represented as a node and each PCI value is a different color. By graph coloring adjacent nodes should have different colors. However, in PCI planning with graph coloring, there are other conditions like preventing confusion. Recently, it has been shown that with physics-inspired graph neural networks, nodes of a graph can be colored. In this paper, a new method for the PCI allocation in an LTE network with a graph neural network is introduced. In this method, the graph neural network is used for PCI allocation to reduce PCI collision, PCI confusion, and mod 3 interference.