System level modeling and evaluation of heterogeneous cellular networks

Abstract

The cumulative impact of co-channel interferers, commonly referred to as aggregate network interference, is one of the main performance limiting factors in today-s mobile cellular networks. Thus, its careful statistical description is decisive for system analysis and design. A system model for interference analysis is required to capture essential network variation effects, such as base station deployment- and signal propagation characteristics. Furthermore it should be simple and tractable so as to enable first-order insights on design fundamentals and rapid exchange of new ideas. Interference modeling has posed a challenge ever since the establishment of traditional macro-cellular deployments. The recent emergence of heterogeneous network topologies complicates matters by contesting many established aspects of time-honored approaches. This thesis presents user-centric system models that enable to investigate scenarios with an asymmetric interference impact. The first approach simplifies the interference analysis in a hexagonal grid setup by distributing the power of the interfering base stations uniformly along a circle. Aggregate interference is modeled by a single Gamma random variable. Its shape- and scale parameter are determined by the network geometry and the fading. The second model extends the circular concept by non-uniform power profiles along the circles. It enables to map substantially large heterogeneous out-of-cell interferer deployments on a well-defined circular structure of nodes. Thereby it considerably reduces complexity while preserving the original interference statistics. The model is complemented by a new finite sum representation for the sum of Gamma random variables with integer-valued shape parameter that allows to identify candidate base stations for user-centric base station collaboration schemes as well as to predict the corresponding rate performance. The third approach applies stochastic geometry to model two-tier heterogeneous cellular networks with respect to the topology of an urban environment. It tackles the asymmetric interference impact by a virtual building approximation and introduces a new signal propagation model that directly relates to the topology characteristics such as building density and -size, which can straightforwardly be extracted from real world data. In the last part of the thesis, the applicability of the introduced models is validated against simulations with the Vienna LTE-Advanced Downlink System Level Simulator. For this purpose, the analytical models are calibrated against results from LTE-A link level simulations. This part also complements the hitherto user-centric investigations with a system-wide performance evaluation, addressing the impact of user clustering as well as small cell density- and isolation. Particular focus is laid on a systematic and reproducible simulation methodology as well as appropriate performance metrics, since conventional figures of merit tend to conceal performance imbalances among users.