High-Throughput Dense Wireless Networks in Unlicensed Spectrum

Abstract

Dense deployment of wireless local area networks (WLANs) will support the prolific data traffic in future wireless networks beyond fifth generation (B5G), either as small cells for cellular data offloading or last-miles for high throughput broadband connectivity. Interference and contention from large numbers of concurrent spectrum sharing nodes, spatial reuse and coexistence of multiple radio access technologies (RATs) are some of the challenges in high density WLANs. This thesis focuses on improving performance of dense WLAN in the presence of interference and contention among densely distributed stations (STAs) and access points (APs).

Interference inherently depends on the distribution of users among the available APs. To that effect, multiuser (MU) based AP association is proposed to associate groups of users to an AP to maximize sum-rate, as opposed to several single-user (SU) frameworks in existence. The MU-AP association problem is solved using a proposed graph-theoretic polynomial time algorithm and a proposed dual ascent method, which require channel knowledge to perform MU-AP association. To avert constant a priori channel information, an alternative framework is proposed to efficiently perform association based on spatial statistics of a network.

For spatial reuse, the physical carrier sensing (PCS) and the energy detection (ED) thresholds determine separation of multiple concurrent transmissions in space. PCS threshold is used by Wireless Fidelity (WiFi) WLAN nodes to detect WiFi-like signals from active transmissions while ED thresholds are used to detect non-WiFi preambles. Using stochastic geometry tools, frameworks are proposed for efficient selection of the PCS and ED thresholds that are network-specific and improve spatial average throughput. With efficient separation of concurrent transmissions, interference is further reduced.

To jointly address the user-AP association and PCS threshold selection problem, a rate-maximizing framework is proposed. A user-AP association solution is first obtained. PCS threshold selection is then optimized based on the achieved user-AP association. Densification, inevitable in future WLANs in improving performance in the presence of large numbers of contending nodes, is critical to delivering the required service level in next generation wireless networks. Analysis of the tradeoffs between densification and overall throughput is provided to determine node densities that achieve maximum rate.