Autonomous neighbor relation detection and handover optimization in LTE

Handovers require serving cells to have knowledge about the existence of neighboring cells. This information is stored in neighbor relation tables (NRTs), which can be automatically generated during a procedure called automatic neighbor relation (ANR) by leveraging measurements carried out by terminals. We show, using simulations, how fast the NRT converges with respect to the number of terminals and their speeds, as well as how incomplete NRTs influence the handover drop rate. By means of a field trial, we demonstrate that ANR works successfully in a real-life environment. In addition, we investigate solutions for problematic handover scenarios and self-optimization of relevant handover parameters based on decentralized approaches. We show that some handover parameters have more impact on network performance than others. Two promising self-optimization methods are discussed, based on either a cost function or a genetic programming technique which searches for a suitable algorithm that determines the parameters.