Driven by the growing popularity of mobile applications, mobile cloud computing has been envisioned as a promising approach to enhance computation capability of mobile devices and reduce the energy consumptions. In this paper, we investigate the problem of multi-user computation offloading for mobile cloud computing under dynamic environment, wherein mobile users become active or inactive dynamically, and the wireless channels for mobile users to offload computation vary randomly. As mobile users are self-interested and selfish in offloading computation tasks to the mobile cloud, we formulate the mobile users' offloading decision process under dynamic environment as a stochastic game. We prove that the formulated stochastic game is equivalent to a weighted potential game which has at least one Nash Equilibrium (NE). We quantify the efficiency of the NE, and further propose a multi-agent stochastic learning algorithm to reach the NE with a guaranteed convergence rate (which is also analytically derived). Finally, we conduct simulations to validate the effectiveness of the proposed algorithm and evaluate its performance under dynamic environment.

Multi-access mobile edge computing (MEC), which enables mobile users (MUs) to offload their computation-workloads to the computation-servers located at the edge of cellular networks via multi-access radio access, has been considered as a promising technique to address the explosively growing computation-intensive applications in mobile Internet services. In this paper, by exploiting non-orthogonal multiple access (NOMA) for improving the efficiency of multi-access radio transmission, we study the NOMA-enabled multi-access MEC. We aim at minimizing the overall-delay of the MUs for finishing their computation requirements, by jointly optimizing the MUs' offloaded workloads and the NOMA transmission-time. Despite the non-convexity of the formulated joint optimization problem, we propose efficient algorithms to find the optimal offloading solution. For the single-MU case, we exploit the layered structure of the problem, and propose an efficient layered algorithm to find the MU's optimal offloading solution that minimizes its overall-delay. For the multi-MU case, we propose a distributed algorithm (in which the MUs individually optimize their respective offloaded workloads) to determine the optimal offloading solution for minimizing the sum of all MUs' overall-delay. Extensive numerical results have been provided to validate the effectiveness of our proposed algorithms and the performance advantage of our NOMA-enabled multi-access MEC in comparison with conventional orthogonal multiple access (OMA) enabled multi-access MEC.

Deep learning has recently been applied to automatically classify the modulation categories of received radio signals without manual experience. However, training deep learning models requires massive volume of data. An insufficient training data will cause serious overfitting problem and degrade the classification accuracy. To cope with small dataset, data augmentation has been widely used in image processing to expand the dataset and improve the robustness of deep learning models. However, in wireless communication areas, the effect of different data augmentation methods on radio modulation classification has not been studied yet. In this paper, we evaluate different data augmentation methods via a state-of-the-art deep learning-based modulation classifier. Based on the characteristics of modulated signals, three augmentation methods are considered, i.e., rotation, flip, and Gaussian noise, which can be applied in both training phase and inference phase of the deep learning algorithm. Numerical results show that all three augmentation methods can improve the classification accuracy. Among which, the rotation augmentation method outperforms the flip method, both of which achieve higher classification accuracy than the Gaussian noise method. Given only 12.5% of training dataset, a joint rotation and flip augmentation policy can achieve even higher classification accuracy than the baseline with initial 100% training dataset without augmentation. Furthermore, with data augmentation, radio modulation categories can be successfully classified using shorter radio samples, leading to a simplified deep learning model and shorter the classification response time.

Multi-access mobile edge computing (MA-MEC) has been envisioned as one of the key approaches for enabling computation-intensive yet delay-sensitive services in future industrial Internet of Things (IoT). In this paper, we exploit non-orthogonal multiple access (NOMA) for the computation offloading in MA-MEC and propose a joint optimization of the multi-access multi-task computation offloading, NOMA transmission, and computation-resource allocation, with the objective of minimizing the total energy consumption of IoT device to complete its tasks subject to the required latency-limit. We firstly focus on a static channel scenario and propose a distributed algorithm to solve the joint optimization problem by identifying the layered structure of the formulated non-convex problem. Furthermore, we consider a dynamic channel scenario in which the channel power gains from the IoT device to the edge-computing servers are time-varying. To tackle with the difficulty due to the huge number of different channel realizations in the dynamic scenario, we propose an online algorithm, which is based on deep reinforcement learning (DRL), to efficiently learn the near-optimal offloading solutions for the time-varying channel realizations. Numerical results are provided to validate our distributed algorithm for the static channel scenario and the DRL based online algorithm for the dynamic channel scenario.

The emerging non-orthogonal multiple access (NOMA), which enables mobile users (MUs) to share same frequency channel simultaneously, has been considered as a spectrum-efficient multiple access scheme to accommodate tremendous traffic growth in future cellular networks. In this paper, we investigate the NOMA downlink relay-transmission, in which the macro base station (BS) first uses NOMA to transmit to a group of relays, and all relays then use NOMA to transmit their respectively received data to an MU. In specific, we propose an optimal power allocation problem for the BS and relays to maximize the overall throughput delivered to the MU. Despite the non-convexity of the problem, we adopt the vertical decomposition and propose a layered-algorithm to efficiently compute the optimal power allocation solution. Numerical results show that the proposed NOMA relay-transmission can increase the throughput up to 30 percent compared with the conventional time division multiple access (TDMA) scheme, and we find that increasing the relays' power capacity can increase the throughput gain of the NOMA relay against the TDMA relay. Furthermore, to improve the throughput under weak channel power gains, we propose a hybrid NOMA (HB-NOMA) relay that adaptively exploits the benefit of NOMA relay and that of the interference-free TDMA relay. By using the throughput provided by the HB-NOMA relay for each individual MU, we study the multi-MUs scenario and investigate the multi-MUs scheduling problem over a long-term period to maximize the overall utility of all MUs. Numerical results demonstrate the performance advantage of the proposed multi-MUs scheduling that adopts the HB-NOMA relay-transmission.

Due to the limited computing resource and battery capability at the mobile devices, the computation-intensive tasks generated by mobile devices can be offloaded to edge servers or cloud for processing. In this paper, we study the multi-user task offloading problem in an end-edge-cloud system, in which all user devices compete for the limited communication and computing resources. Particularly, we first formulate the offloading problem with the goal of maximizing the Quality of Experience (QoE) of the users subject to resource constraints. Since each user focuses on maximizing its own QoE, we reformulate the problem as a Multi-User Task Offloading Game (MUTO-Game). We then identify an important property that for any device, both the communication interference and the degree of computing resource competition can be upper bounded. Based on the property, we further theoretically prove that there exists at least one Nash Equilibrium offloading strategy in the MUTO-Game. We propose the Game-based Decentralized Task Offloading (GDTO) approach to obtain the Nash Equilibrium offloading strategy. Finally, we analyze the upper bound for the convergence time and characterize the performance guarantee of the obtained offloading strategy for the worst case. A series of experimental results are presented, in comparison with both the centralized optimal approach and the approximate approaches.