Inter-RAT Mobility Robustness Optimization in Self-Organizing Networks

The massive growth in mobile data communication requires new more efficient Radio Access Technology (RAT) such as Long Term Evolution (LTE) being deployed on top of legacy mobile communication systems. Inter-RAT handovers are triggered either when the signal level of the serving RAT becomes weak while a sufficiently high signal level is measured from another RAT, or by traffic steering policies for balancing the load among different RATs, for example. Trouble-free operation of inter-RAT handovers requires an optimal setting of the handover parameters which is typically different for each cell and even location. Without knowing the detailed radio propagation conditions, directions and speeds of User Equipments (UEs), network planning can only provide a default setting which needs to be manually optimized during network operation with the aid of drive tests and expert knowledge. This manual optimization requires extensive human intervention which increases Operational Expenses (OPEX) of mobile operators and yields sub-optimal mobility performance due to limited means for more detailed root cause analysis. Therefore, automatic mechanisms have been requested by mobile operators to optimize the inter-RAT handover parameters. This optimization is known as inter-RAT Mobility Robustness Optimization (MRO) which is one of the use cases in Self-Organizing Network (SON).

The technical complexities and requirements on MRO are too difficult to be tackled efficiently and properly by existing manual optimization methods. Considering that mobile networks consist of a high number of cells, the number of handover thresholds to be optimized in a network is significant. Moreover, the intricate dependencies and interactions among the handover thresholds of different neighboring cells make MRO problems even more challenging and complicated. Current optimization methods such as the local search method Simulated Annealing, for example, can be used offline in the network planning phase, however, they cannot be applied online in real-time networks to dynamically react on the changes in the environment and traffic. From that perspective, new optimization methods are needed to address the challenges and limitations imposed by MRO. In this project, several novel and feasible inter-RAT MRO methods have been proposed and analyzed.

New key performance indicators which capture the different types of mobility failure events are proposed by the author of this project for the inter-RAT scenario. An inter-RAT handover is triggered by a dual-threshold measurement event where the first threshold corresponds to the serving cell and the second to the neighboring target cell of another RAT. This dual-threshold measurement event requires a more precise analysis of Too Late Handovers (TLHs). A TLH which is caused by the misconfigured serving cell threshold is distinguished from that which can be resolved by the target cell threshold. Thus, there are two types of TLHs in contrast to the intra-RAT case where a single type of TLH handover exists.

Inter-RAT handover thresholds of currently standardized RATs are configured and optimized cell-specifically. That is, the same handover thresholds are applied by the UEs irrespective of the neighboring handover target cell. The limitations of a cellspecific optimization approach are analyzed and a new cell-group specific optimization approach where the handover thresholds are differentiated with respect to a group of neighboring target cells is proposed. For both cell-specific and cell-group specific optimization approaches, an automatic algorithm is developed to optimize the inter- RAT handover thresholds. In order to analyze the impact of Time-to-Trigger (TTT), which is a time interval affecting the triggering of handovers, the MRO algorithm is extended to allow a joint optimization of handover thresholds and TTT. Based on findings that even cell-group specific parameters cannot resolve all mobility failure events in some cells where radio conditions are not stationary along the cell border, a more advanced location-specific approach is proposed. Unlike cell-based optimization approaches, the handover thresholds are configured and optimized per cell-area and they can be differentiated with respect to neighboring target cells.

Simulative investigations are carried out to evaluate the performance of the different optimization approaches. It has been shown that mobility failure events are rather located in specific cells. Accordingly, the same UEs are probably affected all the time by these mobility failures which leads to high user dissatisfaction. This clearly indicates the need of cell-specific handover thresholds to resolve the mobility problems in some cells. Moreover, it is shown that the optimization of target cell threshold in a cell-group specific manner yields an additional performance improvement compared to cell-specific optimization approach. The joint optimization approach of handover thresholds and TTT has shown advantages only when the handover thresholds are configured cellspecifically rather than cell-group specifically. The mobility failure events that are not resolved by cell-based optimization approaches are mitigated by cell-area based optimization approach.

The investigations and concepts in this project have directly impacted 3rd Generation Partnership Project (3GPP) standard. Several contributions related to cell-specific and cell-group specific optimization approaches have been submitted and adopted by LTE Release (Rel.) 11 standard.