Cloud-based Implementation of an Automatic Pixel-based Neighbour Identification System for Cellular Networks

With the ever-increasing demand for mobile bandwidth, the quantity of deployed cells also increases exponentially, creating heterogeneous networks of cells of varying sizes in order to accommodate the spatial domains that they occupy. This creates issues in Handover performance, where it becomes difficult to manage when and to which base station should a mobile user swap to during a connection without significant losses in performance, due to the large quantity of cells in one space. On the other hand, systems like Automatic Neighbour Relations, present in 4G Long Term Evolution configurations, mostly consider Handover statistics to decide the best neighbours, which is insufficient due to potential problems with signal coverage and load balancing which it cannot answer to. In this paper, an implementation of a cloud-based automated pixel-based neighbour identification system for cellular networks using Amazon Web Services is presented, wherein it generates high quality ranked Neighbour Cell Lists by utilizing cell topology and associated signal strengths in a map defined by a Pixel grid, which itself is built from data provided by the network's Operations Support System. It's completely technology agnostic and it's incorporated as a package in the Metric Software as a Service ecosystem, facilitating network planning assessment by measuring the cells' coverage overlap, and initializing neighbour lists. The proposed method can be used as a starting point for ANR, since it's capable of building a neighbour cell list in some seconds for a new network deployment, which can be then optimised by ANR mechanisms later with an operating network.