Search

Long Range (LoRa) technology, characterized by extended transmission range and low power consumption, is considered a crucial technology for the Industrial Internet of Things (IIoT). However, with increasing network scale, existing LoRa networks have begun to exhibit low energy efficiency (EE) due to severe collisions. Existing methodologies predominantly assume perfect SF orthogonality, neglecting the detrimental effects of interference between signals using different spreading factors (SFs) in the same channel (CH). This study investigates a joint SF, CH, and transmit power (TP) allocation problem with the objective of optimizing the system's uplink EE while fully considering quasiorthogonal SFs, which is an NP-hard optimization problem. To enhance tractability, we decompose the original problem into two subproblems: CH/SF assignment and TP allocation. First, an attention-based deep neural network is designed to generate a near-optimal CH/SF assignment, leveraging both channel state information and channel activity detection-based historical channel occupancy information. Second, given the CH/SF assignment, the optimal TP allocation is determined through the resolution of a convex optimization problem. Experimental results on a real-world LoRa network demonstrate that our proposed method significantly improves the system's uplink EE.

Unmanned aerial vehicles (UAVs) empowered with artificial intelligence (AI) have become a new paradigm for on-demand and intelligent marine monitoring. To enable diverse AI applications, numerous computation-intensive tasks (e.g., image recognition, video processing, path planning, etc.) that cannot be locally executed by UAVs need to be timely and effectively offloaded. Multiple unmanned surface vehicles (USVs) integrated into a USV fleet is appealingly advocated to provide abundant computation resources for computation tasks. In this paper, we propose an energy-efficient USV fleets-assisted collaborative computation offloading scheme for smart maritime services. Specifically, we first propose a collaborative computation offloading framework, where UAVs act as the requesters of computation offloading services, and USV fleets are the helpers. Then, the first-price sealed reverse auction with reserve price is utilized to incentivize USV fleets to assist in executing computation tasks of UAVs, where the reserve price guarantees the satisfied benefits of UAVs. Afterwards, to minimize the energy consumption of executing tasks within the USV fleet under the delay constraint, the joint allocation optimization scheme for computation subtasks and computation capacities is proposed based on the Block Coordinate Descent (BCD) and Alternating Direction Method of Multipliers (ADMM). Simulation results demonstrate that the proposed scheme improves the expected revenue and participation degree of the USV fleet and reduces the overall energy consumption of computation offloading compared to conventional schemes.

A key feature of federated learning (FL) is to preserve the data privacy of end users. However, there still exist potential privacy leakage in exchanging gradients under FL. As a result, recent research often explores the differential privacy (DP) approaches to add noises to the computing results to address privacy concerns with low overheads, which however degrade the model performance. In this paper, we strike the balance of data privacy and efficiency by utilizing the pervasive social connections between users. Specifically, we propose SCFL, a novel Social-aware Clustered Federated Learning scheme, where mutually trusted individuals can freely form a social cluster and aggregate their raw model updates (e.g., gradients) inside each cluster before uploading to the cloud for global aggregation. By mixing model updates in a social group, adversaries can only eavesdrop the social-layer combined results, but not the privacy of individuals. As such, SCFL considerably enhances model utility without sacrificing privacy in a low-cost and highly feasible manner. We unfold the design of SCFL in three steps. i) Stable social cluster formation. Considering users’ heterogeneous training samples and data distributions, we formulate the optimal social cluster formation problem as a federation game and devise a fair revenue allocation mechanism to resist free-riders. ii) Differentiated trust-privacy mapping. For the clusters with low mutual trust, we design a customizable privacy preservation mechanism to adaptively sanitize participants’ model updates depending on social trust degrees. iii) Distributed convergence. A distributed two-sided matching algorithm is devised to attain an optimized disjoint partition with Nash-stable convergence. Experiments on Facebook network and MNIST/CIFAR-10 datasets validate that our SCFL can effectively enhance learning utility, improve user payoff, and enforce customizable privacy protection.

Efficient route planning and traffic scheduling are the key enabling technologies to achieve intelligent transportation systems (ITS). However, the existing schemes that use driving time or driving distance as indicators cannot meet the personalized requirements of vehicles. In addition, the existing research usually optimizes the route planning and traffic scheduling separately, which is difficult to effectively improve the traffic efficiency in practical applications. In this paper, by considering the personalized requirements of vehicles and the target of the traffic manager (TM), a digital-twin (DT) enabled integrated framework is proposed to jointly optimize the personalized route planning and global traffic scheduling. Specifically, based on the collection of traffic data, we first develop a DT architecture for route planning and traffic scheduling. With this DT architecture, we then consider the traffic condition of each road section and design a global traffic scheduling algorithm to help the DT of the TM (DT-TM) set the optimal reward for each road section to motivate and guide the DTs of vehicles (DT-Vs) to select their routes. After that, by considering the preference of each DT-V and the reward of each road section, a personalized route planning algorithm is designed to plan the optimal route for the DT-V. Since the reward of each road section set by the DT-TM and the route selected by each DT-V affect each other, we thus design a joint optimization algorithm based on dynamic iteration to obtain an equilibrium solution for the DT-Vs and the DT-TM. The simulation results show that the proposed framework can schedule the traffic more effectively and bring higher utility to both the vehicles and the TM than the benchmark schemes.