Abstract

It is common knowledge that the fifth generation (5G) of cellular networks will come with drastic transformation in the cellular systems capabilities and will redefine mobile services. 5G (and beyond) systems will be used for human interaction, in addition to person-to-machine and machine-to-machine communications, i.e., every-thing is connected to every-thing. These features will open a whole line of new business opportunities and contribute to the development of the society in many different ways, including developing and building smart cities, enhancing remote health care services, to name a few. However, such services come with an unprecedented growth of mobile traffic, which will lead to heavy challenges and requirements that have not been experienced before. Indeed, the new generations of cellular systems are required to support ultra-low latency services (less than one millisecond), and provide hundred times more data rate and connectivity, all compared to previous generations such as 4G. Moreover, they are expected to be highly secure due to the sensitivity of the transmitted information.

Researchers from both academia and industry have been concerting significant efforts to develop new technologies that aim at enabling the new generation of cellular systems (5G and beyond) to realize their potential. Much emphasis has been put on finding new technologies that enhance the radio access network (RAN) capabilities as RAN is considered to be the bottleneck of cellular networks. Striking a balance between performance and cost has been at the center of the efforts that led to the newly developed technologies, which include non-orthogonal multiple access (NOMA), millimeter wave (mmWave) technology, self-organizing network (SON) and massive multiple-input multiple-output (MIMO). Moreover, physical layer security (PLS) has been praised for being a potential candidate for enforcing transmission security when combined with cryptography techniques.

Although the main concepts of the aforementioned RAN key enabling technologies have been well defined, there are discrepancies between their intended (i.e., vision) performance and the achieved one. In fact, there is still much to do to bridge the gap between what has been promised by such technologies in terms of performance and what they might be able to achieve in real-life scenarios. This motivates us to identify the main reasons behind the aforementioned gaps and try to find ways to reduce such gaps. We first focus on NOMA where the main drawback of existing solutions is related to their poor performance in terms of spectral efficiency and connectivity. Another major drawback of existing NOMA solutions is that transmission rate per user decreases slightly with the number of users, which is a serious issue since future networks are expected to provide high connectivity. To this end, we develop NOMA solutions that could provide three times the achievable rate of existing solutions while maintaining a constant transmission rate per user regardless of the number of connected users.

We then investigate the challenges facing mmWave transmissions. It has been demonstrated that such technology is highly sensitive to blockage, which limits its range of communication. To overcome this obstacle, we develop a beam-codebook based analog beam-steering scheme that achieves near maximum beamforming gain performance. The proposed technique has been tested and verified by real-life measurements performed at Bell Labs.

Another line of research pursued in this thesis is investigating challenges pertaining to SON. It is known that radio access network self-planning is the most complex and sensitive task due to its impact on the cost of network deployment, etc., capital expenditure (CAPEX). To tackle this issue, we propose a comprehensive self-planning solution that provides all the planning parameters at once while guaranteeing that the system is optimally planned. The proposed scheme is compared to existing solutions and its superiority is demonstrated. We finally consider the communication secrecy problem and investigated the potential of employing PLS. Most of the existing PLS schemes are based on unrealistic assumptions, most notably is the assumption of having full knowledge about the whereabouts of the eavesdroppers. To solve this problem, we introduce a radically novel nonlinear precoding technique and a coding strategy that together allow to establish secure communication without any knowledge about the eavesdroppers. Moreover, we prove that it is possible to secure communications while achieving near transmitter-receiver channel capacity (the maximum theoretical rate).