Mobile edge computing (MEC), which pushes cloud computing capability to the network edge, provides a promising platform to support resource-hungry and latency-sensitive mobile applications. By offloading computation workloads to nearby edge servers, mobile users can have their data processed quickly. On the other hand, due to the idea of the Internet of Things (IoT), a vast number of mobile objects are connecting to the Internet at an unprecedented speed. To this end, non-orthogonal multiple access (NOMA) has the potential to address the consequent challenges of spectral efficiency and massive connectivity by allowing multiple users to transmit in the same frequency resource block simultaneously.

In this thesis, we study resource allocation problems in NOMA-enabled computation...[ Read more ]

Mobile edge computing (MEC), which pushes cloud computing capability to the network edge, provides a promising platform to support resource-hungry and latency-sensitive mobile applications. By offloading computation workloads to nearby edge servers, mobile users can have their data processed quickly. On the other hand, due to the idea of the Internet of Things (IoT), a vast number of mobile objects are connecting to the Internet at an unprecedented speed. To this end, non-orthogonal multiple access (NOMA) has the potential to address the consequent challenges of spectral efficiency and massive connectivity by allowing multiple users to transmit in the same frequency resource block simultaneously.

In this thesis, we study resource allocation problems in NOMA-enabled computation offloading under two different scenarios. In the first scenario, we study an energy minimization problem in a NOMA-assisted MEC system by optimizing the users’ power allocation, the server’s computation resource allocation, and subchannel assignment, subject to users’ strict latency constraints. We decompose the formulated non-convex problem into two subproblems. We then solve the power and computation resource allocation subproblem by dual decomposition methods, and solve the subchannel assignment subproblem by branch and bound methods.

In the second scenario, we consider a latency minimization problem for computation offloading with a more flexible application of NOMA technology. In particular, we investigate a hybrid NOMA-orthogonal multiple access (OMA) transmission, which incorporates three offloading methods, namely, hybrid NOMA, pure NOMA, and pure OMA. The offloading-method selection, and the user selection, which determines the roles played by different users in the data transmission, can be appropriately determined based on users’ power allocation and are optimized subject to users’ power and energy budgets. We present a successive convex approximation (SCA)-based algorithm to solve the formulated problem and analytically derive the selection criteria for the offloading strategy.