Multi-link channel modeling and interference characterization for beyond 4G networks

Beyond 4G wireless networks are envisioned to provide ubiquitous connectivity with ultra high capacity density in dense urban environments. For this purpose, an innovative architecture of the multi-beam Base Station (BS) providing high capacity backhaul links to distributed Relay Stations (RSs) which make access links to multiple users in a hierarchical wireless network is introduced recently. However, performance of any wireless network heavily depends on the underlying wireless propagation channel. This is even more important for the network which relies on aggressive frequency reuse (and thus heavy temporal and spectral interference), and unprecedented deployment scenarios (thus unseen spatial signal and interference propagation characteristics). With that perspective, three main components of such hierarchical wireless network are investigated in a top to bottom manner. Two modeling approaches, deterministic and empirical, are used in this thesis to characterize and model the underlying wireless propagation channels. An enhanced Ray-Tracing (RT) tool is used to characterize peculiarities of the multi-beam BS antenna and crucial parameters of relay designs, while two realistic measurement campaigns are conducted at 3.8 GHz in a typical urban microcellular environment to characterize various aspects of multi-user channels. Based on the novel BS architecture, a pathloss model is derived taking into account specicities of the narrow beamwidth antenna, and compared with existing channel models. Impact of the BS antenna on polarized Multiple-Input Multiple-Output (MIMO) aspects is investigated. Moreover, the potential of the array optimization technique in multi-beam antennas is explored to reduce the amount of Co-Channel Interference (CCI) due to excessive frequency reuse in narrow beams as well as Inter-Cell Interference (ICI) due to same frequency beams of the neighbouring cell. The system level simulations are included to show improvements in throughput densities and grade of service (delay and probability of retry). In order to ensure eciency of relay architecture designs in wireless networks, two critical parameters are investigated: (i) the orientation and (ii) the height of the RS. A signicant pathloss reduction of 10 to 15 dB is shown by orienting the relay antenna in the largest power ray direction through RT simulations. Further, the favourable RS height in terms of pathloss is investigated to provide high quality backhaul link. A trade-o between achievable pathloss and interference from the adjacent cell is suggested such that the RS is enabled with high Signalto- Interference Ratio (SINR). Multi-user MIMO (MU-MIMO) techniques consider multiple users in a single cell network to maximize the sum-throughput rather than the link throughput and to manage the resulting multi-user interference. However, the promised leverage of MU-MIMO largely depends upon the decorrelation (i.e. separation) between multi-user links. In this context, rst the validity of the Wide-Sense Stationarity (WSS) assumption is experimentally assessed for non-stationary (users) MIMO propagation channels through the Channel Matrix Collinearity (CMC) metric. Using an appropriate threshold value, typical averaged statioarity distances are found in a range of 5 to 23 m depending upon the scenario under investigation. It is followed with characterization of multi-user separation through three metrics, (i) the Shadow Fading Correlation (SFC), (ii) the CMC and (iii) the Wideband Spectral Divergence (WSD). The empirical evidence illustrates that multiple users can be considered separated in the sense of MU-MIMO when they are physically separated by a distance of 8 to 12 m or an angle of 2 to 6. Furthermore, multiuser separation results are shown roughly irrespective of the investigated metric (or, equivalently, the related signal processing aspect). For each evaluated metric, empirical multi-user separation models are also proposed.