Wireless-powered mobile edge computing (WPMEC) has gained significant attention for enabling energy-sustainable and low-latency Internet of Things (IoT) applications in domains such as smart cities and industrial automation. Heterogeneous computing resources, which encompass devices with diverse computational capabilities, are essential to support adaptive task offloading and resource-efficient service provisioning. However, integrating wireless power transfer (WPT) with mobile edge computing (MEC) in dynamic environments remains challenging, particularly in the efficient coordination of heterogeneous resources. This paper investigates a distributed optimization problem in a wireless-powered heterogeneous MEC network, where base station association, binary offloading decisions, and distributed resource allocation are closely intertwined. We formulate the problem as a mixed-integer nonlinear programming (MINLP) problem and propose a novel decentralized solution framework that decomposes it into two subproblems: deep reinforcement learning is employed to optimize base station selection and task offloading strategies, while convex optimization techniques, including the Lagrangian method, are used to allocate bandwidth and computational resources at the base stations. Simulation results demonstrate that the proposed approach significantly enhances the energy efficiency and computational performance of wireless edge computing compared to existing benchmarks, confirming its practical value for resource-constrained MEC systems.

Edge computing has emerged as a promising paradigm to enable low-latency and high-bandwidth Internet of Things (IoT) applications. However, owing to the intrinsic characteristics of IoT devices, this emerging paradigm still faces bottlenecks such as energy shortages and security vulnerabilities. In this paper, we consider a physical layer security empowered wireless powered mobile edge computing (WP-MEC) system where wireless devices (WDs) can harvest energy from radio frequency signals, and the delivered data is protected against eavesdropping by using the well-known Wyner's wiretap encoding scheme. We focus on maximizing the secure computation rate by jointly optimizing artificial noise power, resource allocation, and offloading decisions for a multi-antenna scenario with wireless devices and a potential eavesdropper. We formulate this sum secure computation rate (SSCR) maximization problem as a mixed-integer programming problem and design an integrated deep reinforcement learning framework consisting of a discrete policy network module, a continuous policy network module, and an optimization module to derive binary offloading decisions, continuous time allocation, and an artificial noise covariance matrix. We evaluate the proposed approach through extensive simulations, and the results demonstrate that our approach can significantly enhance the security, energy efficiency, and computing capacity of wireless edge computing compared with existing works.

The rapid advancement of sixth-generation (6G) networks and artificial intelligence technologies is leading to the emergence of collaborative intelligent transportation systems (C-ITS), which is regarded as an essential trend in the future of transportation. Integrating Internet of Things (IoT) with C-ITS is an efficient solution to provide real-time data collection and status monitoring for vehicles and infrastructures to improve the intelligence and reliability of C-ITS. In order to address the challenges of limited battery energy and low computing power of IoT nodes, integrating wireless power transfer (WPT) with mobile edge computing (MEC) is considered as a promising solution to improve their lifetime and computational capability for IoT nodes. In this paper, we investigate a distributed dynamic computing offloading model for an MEC-enabled C-ITS, where multiple roadside units (RSUs) collaborate to provide offloading services to wireless devices (WDs). We formulate the task offloading and bandwidth resource allocation as a distributed Stackelberg game. The WDs act as leaders, aiming to maximize their computing rate by offloading tasks to RSU or performing local computing. The RSUs act as followers, optimizing their bandwidth allocation based on the WDs’ offloading decisions, thereby improving the overall system computing rate. We prove the existence of a Stackelberg equilibrium (SE) and propose a multi-agent reinforcement learning algorithm to enable WDs to select offloading decisions and help RSUs optimize bandwidth allocation. Numerical simulation results demonstrate that the proposed scheme offers significant performance improvements over existing methods.

Nowadays, the Internet of Things (IoT) acts as a key enabler for smart cities, intelligent transportation systems, precision medicine, smart grids, etc. However, the computing power of IoT nodes is weak, and encryption algorithms cannot be used to ensure the secure transmission of data because of the high complexity. In recent years, the emerging technologies of wireless powered communication network (WPCN) and physical layer security (PLS) are regarded as potential solutions to allow IoT nodes to harvest energy from radio frequency (RF) and ensure secure data delivery. The secrecy rate, which is defined as the difference of main channel capacity and eavesdropping channel capacity, represents the secure data delivery ability. How to maximize the secrecy rate with the rather limited energy for IoT network is a challenging problem. In this paper, we formulate the secrecy rate maximization problem as an optimization problem. We solve this problem by two steps: using one-dimension research to find the optimal energy harvesting duration and find the closed-form solution for transmission power and transmission duration for every node. Compared with the heuristic algorithm, the proposed algorithm can obtain better performance from secrecy rate perspective.

The layered feature of scalable video coding (SVC) offers a sufficient adaptation to unreliable transmission. When network condition drops sharply, enhancement layers will be abandoned, and only base layers are delivered. However, this will cause noticeable visual artifacts due to quality differences between different layers. To alleviate this problem, we novelly introduce a deep learning-based method into video reconstruction phase of scalable bitstreams. A super-resolution motivated recurrent network is proposed to extract and fuse features from both previous high-resolution frames and the current low-resolution frame. To the best of our knowledge, this is the first attempt to improve the performance of scalable bitstreams reconstruction by a specifically designed super-resolution network. By seamlessly integrating the accessible features, significant video quality improvements in terms of PSNR, SSIM, and VMAF are achieved. At the same time, the improvement of overall visual quality stability is apparent under packet lossy networks, indicating both efficiency and robustness of our approach.

Panoramic videos have become more and more popular now. 360-degree videos give users a better experience but put forward a higher request for network at the same time. Many kinds of solutions to meet the high need of bandwidth have been proposed, such as tiled-based transmission, layer-based transmission and so on. Then how to choose the most suitable bitrates to make full use of the network resource is the problem to be solved urgently. In this paper, we propose a method based on reinforcement learning(RL) algorithm to select the bitrates of the region-of-interest adaptively for panoramic videos. We also compare RL algorithm with three traditional algorithms when changing the Field of View(FOV) and find that RL algorithm proves to perform better in minimizing the rebuffer time and providing higher quality video contents under various network conditions.