In future mobile communication systems, the deep integration of communication, sensing, and computation has become a new trend. This paper designs an integrated sensing, computation, and communication (ISCC) system for internet of vehicles (IoV) by leveraging mobile edge computing (MEC) and integrated sensing and communication (ISAC). Specifically, a collaborative sensing data fusion architecture is proposed, where vehicles collaborate with road side unit (RSU) to address the issue of limited sensing range for a single vehicle. Furthermore, the randomness of collaborative sensing tasks arrival in real-world traffic scenarios is simulated, with communication, sensing, and computation being modeled respectively. The joint optimization of wireless and edge computing resources in the ISCC system aims at maximizing the completion rate of collaborative sensing tasks while ensuring system service delay. To address the mixedinteger nonlinear optimization problem, a resource allocation scheme based on deep reinforcement learning (DRL) is developed, which enables adaptive allocation of wireless and edge computing resources through online learning. Simulation results show that agent of DRL can learn a smart resource allocation strategy from the interactive environment, and achieve better performance than conventional resource allocation schemes.
25 September 2023 in IEEE Internet of Things Journal
Although vehicular ad hoc networks (VANETs) significantly enhance traffic convenience, the propagation of erroneous information by malicious vehicles remains a challenging issue. To maintain message reliability, it is crucial to establish a trust management model that can promptly detect malicious vehicles and identify false messages. This article presents a novel trust management model based on blockchain, machine learning, and active detection technology. In the proposed model, we designed a trust evaluation scheme to evaluate the credibility by calculating the direct and indirect trust of the vehicle. To achieve this goal, we use active detection technology to detect indirect trust in vehicles, and then store it in the blockchain. The direct trust of the vehicle is calculated using a Bayesian classifier. The use of active detection technology speeds up the process of filtering out malicious vehicles. Machine learning technology simplifies the complex iterations involved in computing the trust value. Finally, the use of blockchain ensures the consistency and tamper-proofing of the trusted data. The simulation outcomes demonstrate that our approach outperforms the present trust management models.
For millimeter wave massive multiple-input multiple-output systems, the transceiver usually adopts a hybrid precoding structure to reduce complexity and cost, which poses great challenges to the acquisition of channel state information, especially in the case of low signal-to-noise ratio regime. In this article, residual network (ResNet) is employed to address this problem. Firstly, we design a two-stage channel estimation structure to improve the accuracy of channel estimation. Then, we take ResNet as the basic network and integrate UNet structure to build ResNet-UNet model to solve the model degradation problem. Moreover, we use noise2noise algorithm to train the neural network in order to implement the channel estimation in the case that a clean pilot cannot be obtained. Numerical results show that compared with the traditional channel estimation algorithms and deep convolutional neural network algorithm, the proposed approach has higher accuracy and robustness, and achieves performance close to the denoising algorithm using clean targets that is very difficult to be implemented in practical situations.
1 July 2024 in IEEE Transactions on Green Communications and Networking
As a crucial solution to the insufficient computing resources of device in Internet of Vehicles (IoVs) systems, mobile edge computing (MEC) has received widespread attention, especially for tackling delay-sensitive tasks in IoVs. This paper focuses on a multi-roadside units (RSUs) multi-vehicle IoV MEC system with different task delay thresholds. To enhance the system performance in terms of task completion rate, service delay, and energy consumption, a hybrid multi-agent deep reinforcement learning algorithm (HMADRL) based adaptive joint optimization scheme was proposed for computation offloading and resource allocation strategies. Further, a centralized computation offloading and distributed resource allocation framework is designed to reduce communication overhead between multiple agents, and federated learning (FL) technology is used to protect user privacy and accelerate training. The numerical results validate that our scheme improves the performance of IoV MEC system significantly while satisfying system resource and task delay constraints.
To meet the high throughput and low latency requirements of Internet of Vehicle (IoV), future high-frequency bands will inevitably be considered for vehicle-to-everything communication. Significant progress has been made in integrated sensing and communication (ISAC), but severe path loss due to high-frequency bands is unavoidable. Fortunately, reconfigurable intelligent surface (RIS), with its programmability, low power consumption, and cost-effectiveness, etc., has emerged as a promising solution. As IoV aims for seamless interconnection, the explosive growth of data will require incorporating computation into the existing ISAC system to build an integrated-sensing-communication-computation (ISCC) oriented IoV system. This article proposes a beamforming scheme for RIS-assisted IoV for ISCC, focusing on improving the efficiency of RIS-aided road side unit (RSU) receiving computation tasks of vehicle users (VUs). The scheme jointly optimizes the RSU’s beamforming and RIS’s reflection matrix to maximize the IoV system’s sum-rate while ensuring sensing performance, RSU transmit power, RIS element gain, and VUs service quality constraints. To address the complex optimization problem resulting from nonconvex objective functions and constraints, a novel approach based on a dual-critic network in deep reinforcement learning is adopted. This approach effectively alleviates the overestimation problem in Q-value maximization that exists in traditional single-critic networks. Simulation results show that with appropriate parameter settings, the proposed scheme converges to a higher reward value than the classical deep deterministic policy gradient. Moreover, under different numbers of RIS elements and VUs, the proposed scheme outperforms traditional schemes in terms of both the sum-rate and the sensing signal-to-interference-plus-noise ratio.