Interference models for multicellular scenarios

Statistical modeling of interference power is a useful tool to study and investigate the outage probabilities in homogeneous and heterogeneous cellular networks. Also, the ability to analyze and accurately predict the impact of interference via the use of an interference model is essential to improve the average throughput (cell edge throughput). The aggregate interference due to many individual interferers is often very difficult to characterize exactly, especially when lognormal shadowing and path loss is considered. There have been several statistical models for aggregating the interference power yielding in results for homogenous sources where the individual components have the same distribution, but many more detailed studies involve heterogeneous sources. This thesis project builds upon previous statistical models and considers a more realistic and challenging SINR interference model that will characterize and analyze the aggregate interference power. First, the downlink interference power with log normal distribution is modeled using the Fenton-Wilkinson method, where closed form formula for the outage probability is derived when only large scale effects (path loss and lognormal shadowing) are involved. Extending the work to an environment where both large scale effects and small scale effects (fast fading) are taken into consideration, the interference distributions can be observed under the two proposed schemes, which are fitted by log-normal distributions that reduce the complexity. With this general formula, the impacts of path loss, lognormal shadowing and small scale fading on the probability density function (PDF) of the interference power were investigated. It can be seen that fast fading has effect on the interference power and cannot be neglected. The analytical approach is compared with the simulated results which shows that the proposed model provides a flexible way to characterize outage probability. Moreover, the applications of the symmetric alpha stable, Middleton class A and Gaussian Mixture Model in modeling the interference power in homogeneous and heterogeneous networks are demonstrated. Finally, the analytical conditions on the system model parameters are derived and the distributions are determined, on which the statistical properties of interference power can be accurately modeled. This thesis project presents a comprehensive survey of interference models for homogeneous and heterogeneous cellular networks emphasizing their domains of application, illustrated with examples. In addition, it gives insights which can facilitate system performance analysis and interference management.