Designing, modeling and evaluating the next-generation of wireless networks

Modern wireless devices such as tablets and smartphones are pushing the demand for wireless data rates while causing significant stress to existing wireless networks. While successive generations of wireless standards achieve continuous improvement, it is the general understanding of both academic research and the industry that a significant increase in wireless traffic demand can be met only by a dramatically denser spectrum reuse, i.e., by deploying more base stations/access points per square kilometer, coupled with advanced physical (PHY) layer techniques to reduce inter-cell interference. ❧ Enterprise WiFi networks have been deployed following this paradigm for years. As a matter of fact, the density of access points (APs) has increased to a point where inter-cell interference is canceling any additional gains from even denser deployments. At the same time, advanced physical layer techniques have been incorporated into the standards, most notably single-user MIMO in 802.11n and multi-user MIMO in 802.11ac. Cellular networks, unable to satisfy the bandwidth demand of data plans, resort to WiFi offloading, i.e. they deploy WiFi networks to offload traffic from the cellular network. Future cellular network architectures will most likely follow a similar pattern, that is, they will consist of many small cells densely deployed and use advanced physical layer techniques, e.g. massive MIMO. ❧ The goal of this work is to model and evaluate such current and future proposed architectures, ranging from 802.11n MIMO to the Coordinated ""Virtual"" MU-MIMO. We compare their performance and analyze architectural decisions and trade-offs and their impact in various system deployments. We move a step forward from simulating such systems to analytically modeling them so that large scale deployments can be efficiently compared in real-time. Moreover, we propose and analyze MAC layers that enable and take advantage of the underlying PHY layer technologies and simulate the end-to-end system performance. Finally, we introduce new practical network designs that take into account the advances in analog and digital beamforming and outperform, in many occasions, the approaches that are currently found in literature. The proposed hybrid approaches exhibit both practical and less strict deployment requirements compared to coordinated MU-MIMO, but also sizable gains that approach and often surpass their strictly coordinated counterparts and are validated through a real world experiment.