Adaptive Cellular Layout in Self-Organizing Networks using Active Antenna Systems

The rapidly growing demand of capacity by wireless services is challenging the mobile industry with a need of new deployment strategies. Besides, the nature of the spatial and temporal distribution of user traffic has become heterogeneous and fluctuating intermittently. Those challenges are currently tackled by network densification and tighter spatial reuse of radio resources by introducing a heterogeneous deployment of small cells embedded in a macro cell layout. Since user traffic is varying both spatially and temporally, a so called busy hour planning is typically applied where enough small cells are deployed at the corresponding locations to meet the expected capacity demand. This deployment strategy, however, is inefficient as it may leave plenty of network resources under-utilized during non-busy hour, i.e., most of the operation time. Such over-provisioning strategy incurs high capital investment on infrastructure (CAPEX) as well as operating cost (OPEX) for operators. Therefore, optimal would be a network with flexible capacity accommodation by following the dynamics of the traffic situation and evading the inefficiencies and the high cost of the fixed deployment approach. The advent of a revolutionizing base station antenna technology called Active Antenna Systems (AAS) is promising to deliver the required flexibility and dynamic deployment solution desired for adaptive capacity provisioning. Having the active radio frequency (RF) components integrated with the radiating elements, AAS supports advanced beamforming features. With AAS-equipped base station, multiple cell-specific beams can be simultaneously created to densify the cell layout by means of an enhanced form of sectorization. The radiation pattern of each cell-beam can be dynamically adjusted so that a conventional cell, for instance, can be split into two distinct cells, if a high traffic concentration is detected. The traffic in such an area is shared among the new cells and by spatially reusing the frequency spectrum, the cell-splitting (sectorization) doubles the total available radio resources at the cost of an increased co-channel interference between the cells. Despite the AAS capability, the realization of flexible sectorization for dynamic cell layout adaptation poses several challenges. One of the challenges is that the expected performance gain from cell densification can be offset by the ensuing co-channel interference in the system. It is also obvious that a self-organized autonomous management and configuration is needed, if cell deployment must follow the variation of the user traffic over time and space by means of a sectorization procedure. The automated mechanism is desired to enhance the system performance and optimize the user experience by automatically controlling the sectorization process. With such a dynamic adaptation scheme, the self-organizing network (SON) facilities are getting a new dimension in terms of controlling the flexible cell layout changes as the environment including the radio propagation characteristics cannot be assumed stationary any longer. To fully exploit the flexible sectorization feature in three-dimensional space, reliable and realistic propagation models are required which are able to incorporate the dependency of the radio channel characteristics in the elevation domain. Analysis of the complex relationship among various system parameters entails a comprehensive model that properly describes the AAS-sectorization for conducting detailed investigation and carrying out precise evaluation of the ensuing system performance. A novel SON algorithm that automates the AAS-sectorization procedure is developed. The algorithm controls the activation/deactivation of cell-beams enabling the sectorization based cell layout adjustment adaptively. In order to effectively meet the dynamically varying network capacity demand that varies according to the spatial user distribution, the developed SON algorithm monitors the load of the cell, the spatial traffic concentrations and adapts the underlying cell coverage layout by autonomously executing the sectorization either in the horizontal or vertical plane. The SON algorithm specifies various procedures which rely on real time network information collected using actual signal measurement reports from users. The particular capability of the algorithm is evading unforeseen system performance degradation by properly executing the sectorization not only where in the network and when it is needed, but also only if the ensuing co-channel interference does not have adverse impact on the user experience. To guarantee the optimality of the network performance after sectorization, a performance metric that takes both the expectable gain from radio resource and impact of the co-channel interference into account is developed. In order to combat the severity of the inter-cell interference problem that arises with AAS-sectorization between the co-channel operated cells, an interference mitigation scheme is developed in this thesis. The proposed scheme coordinates the data transmission between the co-sited cells by the transmission muting principle. To ensure that the transmission muting is not degrading the overall system performance by blanking more data transmission, a new SON algorithm that controls the optimal usage the proposed scheme is developed. To appropriately characterize the spatial separation of the cell beams being activated with sectorization, a novel propagation shadowing model that incorporates elevation tilt parameter is developed. The new model addresses the deficiencies of the existing tilt-independent shadowing model which inherently assumes a stationary propagation characteristics in the elevation domain. The tilt-dependent shadowing model is able to statistically characterize the elevation channel variability with respect to the tilt configuration settings. Simplified 3D beamforming models and beam pattern synthesis approaches required for fast cell layout adaptation and dynamic configuration of the AAS parameters are developed for the realization of various forms of AAS-based sectorization. Horizontal and vertical sectorization are the two forms of AAS-based sectorization considered in this thesis where two beams are simultaneously created from a single AAS to split the underlying coverage layout in horizontal or vertical domain, respectively. The performance of the developed theoretical AAS-sectorization concepts and models are examined by means of system level simulations considering the Long Term Evolution-Advanced (LTE-A) macro-site deployment within exemplifying scenarios. Simulation results have demonstrated that the SON mechanism is able to follow the different conditions when and where the sectorization delivers superior performance or adversely affects the user experience. Impacts on the performance of existing SON operations, like Mobility Robustness Optimization (MRO), which are relying on stationary cell layout conditions have been studied. Further investigations are carried out in combination with the cell layout changes triggered by the dynamic AAS-based sectorization. The observed results have confirmed that proper coordination is needed between the SON scheme developed for AAS sectorization and the MRO operation to evade unforeseen performance degradation and to ensure a seamless user experience. The technical concepts developed in this thesis further have impacted the $3^\textrm{rd}$ Generation Partnership Project (3GPP) SON for AAS Work Item (WI) discussed in the Radio Access Network (RAN)-3 Work Group (WG). In particular, the observed study results dealing with the interworking of the existing SON features and AAS sectorization have been noted in the standardization work