A Unified Framework for Joint Energy and AoI Optimization via Deep Reinforcement Learning for NOMA MEC-based Networks

System performance and optimization in NOMA mobile edge computing surveillance network using GA and PSO

“…The obtained results demonstrate that the algorithm works well for offloading solutions for the time-varying channel realizations. A similar approach is presented by Zakeri et al in [27] for the scheduling and resource allocation problem of a MEC-assisted RAN. The authors apply the concept of reinforcement learning and implement a deep Q-network (DQN) to provide a solution with low complexity and independent of the lack of some channel information and interference in the system.…”

“…However, the approach based on traditional search algorithms and derivative method has high complexity and is difficult to apply in practical conditions when the number of input parameters is vast. Therefore, to solve the optimizing non-convex problem with many input variables and complex objective function for NOMA-MEC networks, AI techniques have been proposed and attracted much attention from researchers [23,26,27]. Xie et al [23] propose to use a Deep neural network (DNN) to solve the optimization problem of selecting users in the NOMA network and determining the power allocation coefficient for those users.…”

“…The problem of allocating resources in NOMA-MEC networks to ensure latency, energy consumption, and security is attracting much attention. Many solutions have been proposed to optimize the performance of MEC networks, such as using search-based algorithms [10], derivatives methods [11,12], machine learning (ML) [23][24][25][26], or metaheuristic-based algorithms [27,28].…”