Multi-access edge computing (MEC) has been one promising solution to reduce the computation delay of wireless devices. Due to the high spectrum efficiency of non-orthogonal multiple access (NOMA), this paper studies the single-user multi-edge-server MEC system based on downlink NOMA, aiming to minimize task computation delay by jointly optimizing the NOMA-based transmission duration (TD) and workload offloading allocation (WOA) among edge computing servers. This task computation delay minimization (CDM) problem is formulated as a nonconvex optimization problem. To solve the CDM problem efficiently, we decompose it into the sub-problem of determining the optimal WOA with a given TD and the top-problem of optimizing the TD. For the sub-problem, we first derive its some important properties and then design an efficient channel quality ranking based algorithm to obtain the optimal WOA. We solve the top-problem for the static-channel and dynamic-channel scenarios, respectively. For the static-channel scenario, we design an optimal algorithm which only apply once the golden section search method to obtain the optimal TD of first task and directly obtain the optimal offloading solution for any consequently arrived task with different workloads. For the dynamic-channel scenario where the channel qualities from the wireless device to the edge-computing servers are varying, it is critical to quickly determine the current task’s offloading solution under the current channel state and task workload, which is very challenging for the traditional optimization methods. In order to conquer this challenge, we propose the deep reinforcement learning (DRL) based algorithm, which can obtain the near-optimal offloading solution instantly after enough learning. Finally, we validate through simulations the advantages of NOMA over frequency division multiple access (FDMA).

Mobile-edge computing (MEC) and wireless power transfer (WPT) are pivotal for enhancing computational power and battery life in 5G/6G networks. However, their performance declines in remote or disaster-stricken areas due to the lack of access points and energy sources. This paper proposes a wirelessly powered unmanned aerial vehicle enabled MEC (UAV-MEC) system to address this issue, focusing on nodes with ignorable computing capabilities and randomly arriving, size-varying tasks. We aim to maximize the long-term average computation rate under constraints such as UAV coverage, time resources, energy, and task causality, formulating a non-convex problem with dynamic states and complex actions. To solve this problem, we introduce an exploration-enhanced deep reinforcement learning (EDRL) algorithm with a bi-layered structure: the main problem determines the UAV’s flying actions, while the sub-problem allocates time resources given these actions. EDRL employs a deep neural network to analyze real-time UAV positions and task demands, determining optimal flight paths. Upon path determination, an efficient algorithm utilizing bisection and golden section search methods allocates WPT and computational offloading durations. Simulations reveal that EDRL achieves an execution latency of just 11.5 ms in thirty-node networks, outperforming baseline DRL algorithms and predetermined trajectory schemes by 20% and 25% in long-term computation rates, respectively. These results highlight EDRL’s effectiveness and low computational complexity, making it a robust solution for challenging environments.

Non-orthogonal multiple access (NOMA) emerges as a superior technology for enhancing spectral efficiency, reducing latency, and improving connectivity compared to orthogonal multiple access. In NOMA networks, successive interference cancellation (SIC) plays a crucial role in decoding user signals sequentially. The challenge lies in the joint optimization of SIC ordering and power allocation, a task complicated by the factorial nature of ordering combinations. This study introduces an innovative solution, the Attention-based SIC Ordering and Power Allocation (ASOPA) framework, targeting an uplink NOMA network with dynamic SIC ordering. ASOPA aims to maximize weighted proportional fairness by employing deep reinforcement learning, strategically decomposing the problem into two manageable subproblems: SIC ordering optimization and optimal power allocation. We use an attention-based neural network to process real-time channel gains and user weights, determining the SIC decoding order for each user. A baseline network, serving as a mimic model, aids in the reinforcement learning process. Once the SIC ordering is established, the power allocation subproblem transforms into a convex optimization problem, enabling efficient calculation of optimal transmit power for all users. Extensive simulations validate ASOPA’s efficacy, demonstrating a performance closely paralleling the exhaustive method, with over 97% confidence in normalized network utility. Compared to the current state-of-the-art implementation, i.e., Tabu search, ASOPA achieves over 97.5% network utility of Tabu search. Furthermore, ASOPA has two orders of magnitude less execution latency than Tabu search when $N=10$ and even three orders magnitude less execution latency less than Tabu search when $N=20$ . Notably, ASOPA maintains a low execution latency of approximately 50 milliseconds in a ten-user NOMA network, aligning with static SIC ordering algorithms. Furthermore, ASOPA demonstrates superior performance over baseline algorithms besides Tabu search in various NOMA network configurations, including scenarios with imperfect channel state information, multiple base stations, and multiple-antenna setups. These results underscore the robustness and effectiveness of ASOPA, demonstrating its ability to ability to achieve good performance across various NOMA network environments.

This paper addresses long-term energy efficiency in a wireless power transfer-enabled mobile-edge computing (MEC) system, facing challenges from time-varying channels and stochastic task arrivals. We formulate the problem to optimize offloading, power transfer duration, and energy consumption while ensuring queue stability. We propose a novel Lyapunov-guided deep reinforcement learning (LyCNN-DRL) algorithm to efficiently solve the long-term mixed integer non-linear programming problem without prior knowledge of future conditions. The approach decomposes the problem into resource allocation and binary offloading components, using a convolutional neural network for near-optimal offloading decisions and the Lagrange dual function for optimal resource allocation. Extensive simulations show that LyCNN-DRL outperforms benchmark algorithms in energy efficiency and latency, achieving over 97% of the optimal utility while reducing execution latency to approximately 50 milliseconds in ten-WD networks. Additionally, we derive the trade-off between energy efficiency and queue length as [O(1/V),O(V)], where V is the Lyapunov control parameter.

Wireless power transfer enhanced mobile edge computing (WPT-MEC) has emerged as a key technology to support low-latency and energy-efficient computation in wireless networks. With increasing network density, multi-base-station architectures emerge where wireless devices (WDs) offload tasks to distributed base stations (BSs), creating challenges in maintaining quality-of-service fairness during complex resource coordination in multi-BS WPT-MEC networks. To address these challenges, we investigate a non-orthogonal multiple access (NOMA)-enhanced WPT-MEC network comprising multiple WDs and BSs with finite computational capacities. For ensuring fairness, we formulate a max-min problem to maximize the minimum task computation amount by jointly optimizing offloading decisions, NOMA decoding orders, offloading powers and time resource allocation, which results in a challenging mixed integer, sequence and nonlinear programming (MISNLP). To tackle this problem, we propose a two-stage distributed multi-agent algorithm. In the first stage, each BS agent generates offloading preferences based on partial observations, guiding WDs’ offloading decisions. In the second stage, given these offloading decisions, we develop an efficient convex-based algorithm to solve the per-BS resource allocation subproblem, jointly optimizing NOMA decoding order, offloading powers and time resource allocation. For effective training, we leverage off-policy training and the centralized training with decentralized execution (CTDE) paradigm with two key innovations: (1) a convex-based critic that evaluates the joint action without bias, and (2) a counterfactual baseline that isolates individual agent credit assignment. The proposed C3MA algorithm achieves six times faster convergence and at least 20% performance improvement when serving more than 20 WDs, compared with existing multi-agent schemes, while maintaining a near-optimal Jain’s fairness index of 0.97. Moreover, it sustains an ultra-low execution delay below 5 milliseconds even with 40 WDs, confirming its efficiency and scalability.

Unmanned Aerial Vehicle (UAV)-enabled wirelessly powered Mobile Edge Computing (MEC) is emerging as a powerful technology for boosting computational capability and energy supplementation in Internet of Things (IoT). This work addresses the long-term task completion ratio maximization problem in UAV-enabled wirelessly powered MEC systems. Besides the large number of optimization parameters, the environment can only be partially observed as the UAVs cannot cover the whole network area. Then, it is very challenging to obtain good solutions due to the lack of global information. We introduce a novel distributed Multi-Agent Deep Reinforcement Learning (MADRL) framework for optimizing UAVs’ actions and resource allocation, considering the constraints of tasks that vary in size, arrival times, and required computation completion time. To decouple the complicated parameters, we divide the problem into two manageable subproblems—UAVs’ action decision and resource allocation under a given UAV’s action. We employ a distributed Deep Reinforcement Learning (DRL) scheme for the former subproblem to cope with the partially observable nature. By revealing some important properties of the later subproblem, we design an efficient two-stage optimal algorithm to minimize the total consumed energy of nodes while maximizing the task-completing number. Extensive simulations validate the effectiveness of the proposed framework, achieving over a 50% improvement in task completion ratio compared to baseline schemes in some scenarios.