Efficient radio resource management in integrated terrestrial and non-terrestrial networks

Birabwa, Denise Joanitah

Efficient radio resource management in integrated terrestrial and non-terrestrial networks

Thesis / Dissertation

2024

Publisher

University of Cape Town

Department

Department of Electrical Engineering

Faculty

Faculty of Engineering and the Built Environment

Abstract

The beyond 5G (B5G) networks are envisaged to provide terra bps data rates and ubiquitous and unlimited wireless coverage. However, the terrestrial deployment of 5G networks poses a limitation in achieving a truly ubiquitous and seamlessly connected network. To this end, it has been proposed to integrate terrestrial networks (TNs) with non-terrestrial networks (NTNs), such as satellite communications, high-altitude platforms and low-altitude platforms. NTNs are characterised by wide coverage and less vulnerability to physical attacks and natural disasters, and hence, will complement TNs in providing ubiquitous wireless connectivity and enhanced broadband services to unserved and underserved areas. In addition, NTNs will provide network resilience to physical attacks and natural disasters, improve the quality of service (QoS) for overloaded TNs, and enhance service continuity for moving platforms. Consequently, the B5G network will be an integrated terrestrial and nonterrestrial network (ITNTN) consisting of multiple radio access networks (RANs) coexisting to provide radio access to multi-mode user equipment. This thesis addresses the problem of efficient user association and resource allocation in the ITNTN. The RANs in the ITNTN have different capabilities and limitations in meeting the envisioned B5G contrasting user requirements such as throughput, latency, and mobility. Therefore, determining an optimal association and resource allocation scheme that maps the heterogeneous users to the appropriate RANs while at the same time maximising resource utilisation, and providing the required user QoS, is rigorous and complex. Besides, such coexistence of the different RANs implies an increase in the number of wireless access nodes and thus raises a justifiable concern over the drastic increase in energy consumption and carbon emission expected to ensue. Accordingly, there is a need to develop efficient radio resource management (RRM) algorithms for the ITNTN that consider not only the heterogeneity in user QoS requirements but also the uniqueness of the different RANs in meeting these demands. To this end, this research aims at developing efficient RRM schemes that achieve a spectrum-efficient and energy-efficient ITNTN while minimising mobility-induced handoffs. RRM takes the form of user association and resource allocation in this work. First, the research formulates the user association and resource allocation problem in the ITNTN as a multi-objective optimisation problem (MOOP) that jointly maximises the total data rate of the ITNTN while minimising the probability of mobility-induced handoffs. The problem is subjected to constraints on the resource budget and minimum user QoS requirements in terms of data rate. Moreover, the problem is formulated to allow differentiated service provisioning and priority-based user association and resource allocation, thus prioritising mission-critical users' service provisioning. The weighted sum method is adopted to simplify and transform the MOOP into a single-objective optimisation problem (SOOP). The SOOP's complexity is reduced by decomposing it into two sub-problems: the user association sub-problem and the resource distribution sub-problem. A service-aware greedy heuristic algorithm is proposed to solve the user association sub-problem and its performance compared to the serviceunaware scheme. Simulation results reveal that the service-aware algorithm achieves higher overall network spectrum efficiency (SE), user acceptance ratio (AR), and lower handoff probability. Furthermore, the resource distribution sub-problem is reformulated into a waterfilling problem and solved utilising CVXPY, consequently analysing the effect of distributing the unallocated basic bandwidth units to the associated users. Second, since the greedy heuristic solution to the user association sub-problem does not ordinarily produce an optimal solution, the work further proposes a polynomial-time solution based on the genetic algorithm (GA). The performance of the GA is evaluated by comparing it to the ILP solution, the greedy heuristic solution, and the random user association (RUA) algorithm. Simulation results reveal that the GA outperforms all algorithms in terms of SE and user acceptance ratio. Moreover, the GA achieves a handoff probability of zero, unlike the RUA algorithm. Third, the proposed greedy heuristic and GA solutions to the user association subproblem utilise a central node that requires nearly-complete information, which may not be available in real-time. Therefore, this research further proposes a centralised training and distributed execution multi-agent duelling double deep Q network (MA3DQN) solution that facilitates real-time decision-making. Each user collects the channel state and access node loading information in this approach and makes an association decision that considers its quality of service requirements. This section of the thesis further adopts the effective capacity theorem to guarantee the delay QoS requirements for mission-critical users. The MA3DQN's performance is validated through comparison with the GA, the ILP solution, a heuristic approximation-based solution, the greedy approach, and the RUA algorithm. Moreover, the multi-agent deep Q network (MADQN) solution is also simulated as an additional benchmark algorithm. Simulation results reveal that as the number of users in the network increases, the acquired data rate of the MA3DQN is within 0.48% and 0.42% of that achieved by the GA and ILP, respectively, and outperforms all other algorithms. Notably, the proposed MA3DQN algorithm presents the best running time, attaining a gain of 99.9% over the GA algorithm, which performs the poorest among the algorithms characterised by polynomial worst-case time complexity. Besides, the MA3DQN approach maintains a handoff probability of zero, unlike the approximation-based, and RUA solutions. Lastly, the thesis presents a weighted sum SOOP that maximises the energy efficiency (EE) of the ITNTN while simultaneously minimising the mobility-induced handoff probability. The formulated problem is a non-convex and mixed integer non-linear programming problem whose complexity is reduced through decomposition into two sub-problems: the user association and resource allocation (UARA) problem and the power allocation (PA) problem. The equivalent UARA problem maximises the total network data rate and minimises the mobility-induced handoffs, thus can be solved by the GA or the MA3DQN already discussed. On the other hand, the PA problem is a fractional programming problem (FPP) that is simplified through transformation into a weighted sum SOOP and solved using the particle swarm optimisation (PSO) approach. Simulation results reveal that depending on the value of the weighting factor in the SOOP, the particle swarm optimisation power allocation (PSOPA) algorithm either minimises power consumption, thereby maximising the EE of the ITNTN, or maximises the total network data rate. When the power consumption is minimised, the PSOPA outperforms the equal power allocation (EPA) technique in EE. On the other hand, when the achieved network data rate is maximised, the PSOPA's achieved data rate approaches the upper bound set by the EPA algorithm. Unlike many recent proposals in the literature on user association and resource allocation in the ITNTN, the algorithms proposed in this thesis consider the heterogeneity in user traffic types. Moreover, the capabilities and limitations of the different networks in meeting the contrasting user demands are considered. For instance, to the best of our knowledge, no work in literature prioritises the wide coverage non-terrestrial networks over terrestrial networks for long-distance and highly mobile applications. Such a consideration reduces the handoff rate, ultimately reducing the probability of handoff failure, delays, and communication overheads.

Keywords

Electrical Engineering

Reference:

Collections

PhD / Doctoral

Full item page