In current wireless communication systems, demands for extremely high data rates, along with spectrum scarcity at the microwave bands, make the millimeter wave (mmWave) band very appealing to provide these extremely high data rates even for a massive number of wireless devices. MmWave communications exhibit severe attenuation, vulnerability to obstacles (called blockage), and sparse-scattering environments. Moreover, mmWave signals have small wavelengths that allow the incorporation of many antenna elements at the current size of radio chips. This leads to high directivity gains both at the transmitter and at the receiver, directional communications, and, more importantly, possible noise-limited operations as opposed to microwave networks that are mostly interference-limited.
These fundamental differences between mmWave networks and legacy communication technologies challenge the classical design constraints, objectives, and available degrees of freedom. The natural consequence is the necessity of revisiting most of the medium access control (MAC) layer design principles for mmWave networks, which have so far received less attention in the literature than physical layer and propagation issues. To address this important research gap, this thesis investigates the fundamental MAC layer performance metrics, including coverage, fairness, connection robustness, collision probability, per-link throughput, area spectral efficiency, and delay. The original analysis proposed in this thesis suggests novel insights as to the solutions for many MAC layer issues such as resource allocation, interference management, random access, mobility management, and synchronization in future mmWave networks.
A first thread of the thesis focuses on the fundamental performance analysis and mathematical abstraction of mmWave wireless networks to characterize their differences from conventional wireless networks, i.e., high directivity, line-of-sight communications, and occurrence of deafness (misalignment between transmitters and receivers). A mathematical framework to investigate the impact of beam training (alignment) overhead on the throughput is established, which leads to identify a new alignment-throughput tradeoff in mmWave networks. A novel blockage model that captures the angular correlation of line-of-sight conditions using a new notion of "coherence angle" is proposed. The coverage and delay of directional cell discovery are evaluated, and an optimization approach to maximize long-term throughput of users with fairness guarantees is proposed. In addition, this thesis develops a tractable approach to derive the collision probability, as a function of density of the transmitters, transmission power, density and size of the obstacles, operating beamwidth, and sensitivity of the receiver, among the main parameters. The collision probability allows deriving closed-form expressions for the per-link and network throughput of mmWave networks, and thereby identifying that, contrary to mainstream belief, these networks may exhibit a non-negligible transitional behavior of interference from a noise-limited to an interference-limited regime.
The second thread of the thesis builds on the previous fundamental performance analysis and modeling to establish new, efficient MAC protocols. The derived collision probability is used to evaluate per-link throughput, area spectral efficiency, and delay performance of common MAC protocols such as TDMA and slotted ALOHA, and to provide a fundamental comparison between pros and cons of contention-free and contention-based MAC protocols. The results suggest the use of on-demand interference management strategy for future mmWave cellular networks and collision-aware hybrid MAC protocols for mmWave ad hoc networks to reliably deliver messages without sacrificing throughput and delay performance. Moreover, the transitional behavior, together with significant mismatch between transmission rates of control and data messages, imposes the development of new hybrid proactive and reactive control plane architecture. This thesis identifies the prolonged backoff time problem, which happens in mmWave networks due to blockage and deafness, and proposes a new collision notification signal to solve this problem. Motivated by the significant mismatch between coverage of the control and data planes along with delay analysis of directional cell search, a novel two-step synchronization procedure is proposed for mmWave cellular networks. Also, the impact of relaying and multi-hop communication to provide reliable mmWave connections, to alleviate frequent handovers, and to reduce the beam training overhead is investigated.
The investigations of this thesis aim to demystify MAC layer performance of mmWave networks and to show the availability of many new degrees of freedom to improve the network performance, e.g., in terms of area spectral efficiency, energy efficiency, robustness, delay, coverage, and uniform quality of service provisioning. The results reveal many special behaviors of mmWave networks that are largely ignored in design approach of the current mmWave networks. Given that the standardization of mmWave wireless cellular networks has not started as yet, and that existing standards of mmWave ad hoc networks are highly sub-optimal, the results of this thesis will provide fundamental design guidelines that have the potential to be very useful for future mmWave standardizations.
Stockholm: KTH Royal Institute of Technology, 2015. , xvi, 23 p.