Multi-tier cellular networks are a cost-effective solution for capacity enhancement in urban
scenarios. In these networks, effective handover schemes are required to assign users to the
most adequate layer. In this paper, a data-driven self-tuning algorithm for user steering is
proposed to improve the overall Quality of Experience (QoE) in multi-carrier Long Term
Evolution (TE) networks. Unlike classical approaches, user steering is achieved by changing
Reference Signal Received Quality (RSRQ) based inter-frequency handover margins. To drive
the tuning process, a novel indicator showing throughput changes in the vicinity of handovers is
derived from connection traces. Method assessment is carried out in a dynamic system-level
LTE simulator implementing a real multi-carrier scenario. Results show that the proposed
algorithm significantly improves QoE figures obtained with a classical inter-frequency
handover scheme based on Reference Signal Received Power (RSRP) measurements.
