Integrated terrestrial and non-terrestrial power internet of things (IPIoT) has emerged as a paradigm shift to three-dimensional vertical communication networks for power systems in the 6G era. Computation offloading plays key roles in enabling real-time data processing and analysis for electric services. However, computation offloading in IPIoT still faces challenges of coupling between task offloading and computation resource allocation, resource heterogeneity and dynamics, and degraded model training caused by electromagnetic interference (EMI). In this article, we propose an asynchronous federated deep reinforcement learning (AFDRL)-based computation offloading framework for IPIoT, where models are uploaded asynchronously for federated averaging to relieve network congestion and improve global model training. Then, we propose Asynchronous fedeRated deep reinforcemenT learnIng-baSed low-laTency computation offloading algorithm (ARTIST) to realize low-latency computation offloading through joint optimization of task offloading and computation resource allocation. Particularly, ARTIST adopts EMI-aware federated set determination to remove aberrant local models from federated averaging and improve training accuracy. Next, a case study is developed to validate the excellent performance of ARTIST in reducing task offloading and total queuing delays.
21 June 2024 in IEEE Transactions on Green Communications and Networking
The low-carbon and efficient operation of smart parks requires high-precision and real-time energy management model training. Multi-mode power internet of things (PIoT) consisting of open radio access networks (O-RAN) and power line communications (PLC) can effectively improve the model training performance. However, the negative effects of network threats, such as model inversion attacks, cannot be neglected. To solve this problem, we propose a diFferential pRivacy-aware gEnErative aDversarial netwOrk-assisted resource scheduling algorithM (FREEDOM). Firstly, we integrate a differential privacy mechanism with the energy management model training process and the related system model. Then, a joint resource scheduling optimization problem is constructed, the goal of which is to minimize the weighted sum of the global loss function, total energy consumption, and differential privacy cost under the long-term differential privacy constraint. The original problem is converted based on virtual queue theory and addressed by the FREEDOM. FREEDOM leverages a deep Q-learning network (DQN) to learn the resource scheduling strategy via differential privacy awareness. It improves optimization and convergence performances with the assistance of generative adversarial network (GAN). Simulation results show that FREEDOM can achieve excellent performances of global loss function, total energy consumption, differential privacy cost, and privacy preservation.
With the development of the smart grid, massive electric Internet of Things (EIoT) devices are deployed to collect data and offload them to edge servers for processing. However, the task of offloading optimization still faces several challenges, such as the differentiated quality of service (QoS) requirements, decision coupling among multiple devices, and the impact of electromagnetic interference. In this paper, a low-complexity delay and energy-efficiency-balanced task offloading algorithm based on many-to-one two-sided matching is proposed for an EIoT. The proposed algorithm shows its novelty in the dynamic tradeoff between energy efficiency and delay as well as in low-complexity and stable task offloading. Specifically, we firstly formulate the minimization problem of weighted difference between delay and energy efficiency to realize the joint optimization of differentiated QoS requirements. Then, the joint optimization problem is transformed into a many-to-one two-sided matching problem. Through continuous iteration, a stable matching between devices and servers is established to cope with decision coupling among multiple devices. Finally, the effectiveness of the proposed algorithm is validated through simulations. Compared with two advanced algorithms, the weighted difference between the energy efficiency and delay of the proposed algorithm is increased by 29.01% and 45.65% when the number of devices is 120, and is increased by 11.57% and 22.25% when the number of gateways is 16.
Quantum-classical hybrid encryption has elevated security levels of Internet of Things (IoT)-enabled distributed power control. However, the decoupled design of communication and power systems introduces substantial technical challenges. Outdated or tampered state data propagating from communication to power domains trigger cascading control failures, as demonstrated in the 2015 Ukraine grid cyberattack. This motivates us to investigate the fundamental question of how to achieve power-communication collaborative control optimization by proposing a cross-domain gap driven optimization framework of integrated sensing, transmission and control. Our novel framework addresses these limitations through three synergistic innovations: (i) accurate false data injection (FDI) attack sensing based on temporal correlation, (ii) secure data transmission based on hierarchical cooperative learning, and (iii) trust and delay-weighted consensus control. Its superiority is validated through simulations across diverse attack scenarios, grid topologies, and network scales. It exhibits strong scalability and robustness in large-scale complex grids. Key finding revels that blocking propagation paths of outdated and tampered consensus variables effectively mitigates control deviations and oscillations.
Precise energy management in distribution power system requires high‐precision time synchronization among large‐scale deployed devices. Multiple clock sources‐based time synchronization possesses advantages of reliability, high precision, and robustness, but still faces several challenges such as coupling between time synchronization error and delay, as well as different timescales between clock source and clock weight optimization. In this paper, a multi‐clock source time synchronization model is constructed and a problem is formulated to minimize the synchronization error and delay through jointly optimizing large‐timescale clock source selection and small‐timescale weight selection. A reinforcement learning‐based multi‐timescale multi‐clock source time synchronization algorithm named RL‐M 2 is proposed to solve the formulated problem from a learning perspective. Besides, a lossless switching method is proposed to address the switching problem for multiple clock sources. Simulation results demonstrate the superior performance of RL‐M 2 and the lossless switching method in time synchronization delay and error.
29 October 2023 in Digital Communications and Networks