Abstract:Multi-access edge computing has been considered as a promising solution for enabling computation-intensive yet latency-sensitive applications at resource-constrained wireless devices (WDs). To improve the spectrum efficiency for multi-WD computation offloading, this paper considers nonorthogonal multiple access (NOMA) assisted two-tier multiaccess edge computing scenario, which exploits the computation resources of both the edge servers (ESs) and the cloudlet server (CS) deployed at different tiers. In particular, the WDs can offload partial workloads to different ESs simultaneously via NOMA, and the ESs can form a NOMA-group to further offload partially received workloads to the CS for processing. We investigate the total energy consumption minimization problem by jointly optimizing the two-tier offloading decisions, the NOMA transmission duration, and the computation resource allocation. Due to the successive interference cancellation in the NOMA and the coupling effect in two-tier offloading, the formulated optimization problem is strictly non-convex. To address this difficulty, we exploit the hierarchical relationship among the joint optimization variables, and then propose a hybrid deep reinforcement learning (HDRL) algorithm to learn two policies that determine the coupled variables, i.e., the ESs' offloading decisions and the NOMA transmission duration, respectively. Then, the remaining decision variables can be jointly optimized by using the convex optimization methods directly based on the results provided by the HDRL algorithm. Specifically, the HDRL algorithm that uses different policies to determine the coupled variables can converge faster than the existing solutions that learn a single policy to determine all variables. Experimental results are provided to validate the performance of our proposed HDRL algorithm in comparison with two other learning-based algorithms.

DRL-Assisted Energy Minimization for NOMA-Based Dynamic Multi-User Multi-Access MEC Networks

Two-Tier Multi-Access Partial Computation Offloading Via NOMA: A Hybrid Deep Learning Approach for Energy Minimization

Non-orthogonal Multiple Access Assisted Mobile Edge Computing Via Device-to-Device Communications

Learning-Assisted Energy Minimization for MEC Systems With Noncompletely Overlapping NOMA

Learning-Assisted Partial Offloading for Dynamic NOMA-MEC Systems with Imperfect SIC and Reconfiguration Energy Cost

Computation Offloading and Resource Allocation in NOMA-MEC: A Deep Reinforcement Learning Approach

Efficient Resource Allocation for NOMA-MEC System in Ultra-dense Network: A Mean Field Game Approach

Energy-efficient User Clustering and Resource Management for NOMA Based MEC Systems

Energy Saving Computation Offloading for Dynamic CR-MEC Systems with NOMA Via Decomposition Based Soft Actor–critic

Task Offloading, Load Balancing, and Resource Allocation in MEC Networks

Deep Reinforcement Learning Based Computation Offloading and Resource Allocation for MEC

Energy-Delay Tradeoff in Device-Assisted NOMA MEC Systems: A Matching-Based Algorithm

Latency Optimization for Multi-user NOMA-MEC Offloading Using Reinforcement Learning

DRL-Assisted Resource Allocation for NOMA-MEC Offloading with Hybrid SIC

Joint Resource Allocation and Task Offloading for Hybrid NOMA-assisted MEC Network with Network Slicing.

NOMA-Based Multi-User Mobile Edge Computation Offloading via Cooperative Multi-Agent Deep Reinforcement Learning

NOMA Assisted Multi-Task Multi-Access Mobile Edge Computing via Deep Reinforcement Learning for Industrial Internet of Things

Efficient Offloading for Minimizing Task Computation Delay of NOMA-Based Multiaccess Edge Computing

Energy-Efficient Collaborative Multi-Access Edge Computing Via Deep Reinforcement Learning

Resource Allocation for NOMA-MEC Systems in Ultra-Dense Networks: A Learning Aided Mean-Field Game Approach

Secure and Energy-Efficient Offloading and Resource Allocation in a NOMA-Based MEC Network