UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Resource allocation algorithms and preamble design for massive IoT systems Mostafa, Ahmed Elhamy


With the proliferation of the Internet of things (IoT) applications, it becomes essential for wireless cellular networks to support energy-efficient communication for an increasing number of IoT devices. In this thesis, we develop resource allocation algorithms and propose novel preamble design for enhancing the connection density in IoT systems. First, we propose a non-orthogonal multiple access (NOMA) scheme for narrowband IoT (NB-IoT) systems. This scheme allows multiple IoT devices to simultaneously access either one subcarrier (single-tone mode) or a bond of contiguous subcarriers (multi-tone mode). We formulate joint subcarrier and power allocation problems for both modes to maximize the connection density. The formulated problems are nonconvex mixed integer programming problems. We optimally solve the formulated problems using mixed integer linear programming transformation and difference of convex programming. We also propose low-complexity algorithms to solve both problems in a suboptimal manner. Second, we propose a communication mode selection scheme for IoT devices that can communicate using either active transmission or energy-efficient short-range backscattering. In the active transmission mode, the IoT devices can transmit data using power-domain NOMA. In the backscattering mode, nearby user equipment (UE) devices are used as relays that receive the backscattered signals from the IoT devices and forward them to the base station (BS). We formulate a connection density maximization problem to select the communication mode for each IoT device. The optimal algorithm, which solves the formulated binary integer programming problem, incurs exponential computational complexity. Hence, we propose a low-complexity suboptimal algorithm to solve the problem. Third, we propose a larger set of random access preambles by considering all possible combinations of aggregating two Zadoff-Chu preamble sequences. Decoding the aggregate preambles is challenging because the receiver needs to decode two preambles where each one is allocated half of the transmit power. We propose two receiver architectures for preamble decoding. The first architecture only requires minor changes to the conventional preamble receiver architecture. The second architecture exploits a deep neural network (DNN). Simulation results demonstrate that the proposed schemes in this thesis can enhance the capability of wireless cellular networks to support a higher connection density in IoT systems.

Item Citations and Data


Attribution-NonCommercial-NoDerivatives 4.0 International