Abstract
The ongoing development of mobile cellular networks, supporting a wide range of applications and services with high data-rate and ubiquitous connectivity requirements, has resulted in a considerable increase in capacity demand. With the advent of next generation of mobile cellular networks, it is expected that the capacity demand to exceed far beyond the supply. This imminent capacity shortage has introduced impetus to identify practical solutions towards a more efficient utilisation of the spectrum. In this respect, various approaches under the umbrella of Spectrum Sharing (SS) have been explored. However, the incorporated techniques, along with some sets of strict assumptions, have imposed conservatively broad boundaries, known as Exclusion Zones (EZs) to ensure interference protection, irrespective of the actual spatial-temporal utilisation of the spectrum. This has resulted in limited achievable gains, and consequently the conventional approaches are identified as inefficient for the real-world deployment. In fact, for the SS to be efficient and practically viable, there is a need for a paradigm shift towards a more dynamic mechanisms with a level of live spectrum usage awareness in the network, which can efficiently identify interference-free Spectrum Opportunities (SOPs) for sharing, and hence, the shortcomings and constraints of the conventional SS approaches can be mitigated. In this context, the aim of this thesis is to investigate a novel and efficient SS mechanism, in which Radio Environment Map (REM) technique as an enabler is applied. The REM captures near real-time spectrum utilisation in the network in temporal and spatial domains pro-actively, resulting in increased SOPs for Sharing. A comprehensive literature survey of the SS is provided in this thesis. The concepts, various authorisation regimes, along with their specifications and requirements are discussed. Moreover, the potential sharing deployment scenarios, as well as the use cases in which the mobile cellular networks can gain benefit from SS are pointed out. Further, having a robust view of State-Of-The-Art (SOTA) coordination protocols (i.e., centralised and sensing based approaches) and enabling techniques, the associated advantages, as well as the major shortcomings and challenges are investigated. This is followed by providing an in-depth insight into the SOTA proposals, approaches, the respective achieved gains, and the necessity for the enhanced/new techniques. Consequently four techniques, namely Inter-Operator Inter-Cell Interference (IO-ICI), the Sensing, Coordinated Beamforming, and REM identified as promising dimensions that can be substantially enhanced/applied in SS. Focusing on the adoption of REM technique, a SS mechanism is proposed which exploits Received Signal Strength (RSS) along with spatial interpolation techniques to model temporal and spatial map of SOPs in the downlink of Long Term Evolution-Advanced (LTE-A), in a dynamic manner, subject to update rate in the order of LTE-A time frame. The investigation is performed over the two well-known and distinctive spectrum sharing schemes; (1) Inter-Operator Spectrum Sharing (IOSS), and (2) Licensed Shared Access (LSA) 1 . For the scheme (1), the sharing players comprise two large-scale independently deployed Mobile Network Operators (MNOs), over the two standardised multi-MNO deployment topologies; non-collocated and collocated, in urban environment. The simulations are performed with high data rate real-time video streaming traffic traces. The simulation results are compared to the two SOTA approaches (i.e., centralised and sensing based approaches), as well as the LTE-A baseline. The simulation results demonstrate that the proposed REM-based sharing mechanism results in 23% improvement in Spectrum utilisation efficiency, 37.5% average system throughput, with respect to the baseline LTE-A, where the SS is not applied. Moreover, it is observed that the REM-based approach outperforms the two considered SOTA approaches. The cost of overhead, and computational complexity of implementation are found negligible. In addition to IOSS, for the scheme (2) (i.e., the LSA), an arbitrary LSA incumbent as a worst case scenario (when no priori information is given) is considered. Through the the simulation results it is shown that the proposed approach reduces the size of EZs from considerable number of cells to a fewer numbers. The transmit power level does not need to to be reduced in majority of the cells in the network, and thus, the LSA bands can be utilised in a more dynamic manner. As a result, the overall system throughput is significantly increased with respect to the SOTA approach by 80%. However, this gain is subject to fast and reliable interface between two networks to allocate sufficient time for band evacuation.