Perception plays a vital role in autonomous driving as it serves as a prerequisite for downstream planning and decision tasks. The existing research has mainly focused on developing vehicle-side perception models using a single type of sensors. However, relying solely on one type of on-board sensors to perceive the surrounding environment leads to perceptual deficiencies owing to inherent characteristics and sensor sparsity. To address this bottleneck, we propose ViT-FuseNet, a novel vehicle-infrastructure cooperative perception framework that utilizes a Vision Transformer to fuse feature maps extracted from LiDAR and camera data. The key component is a multimodal fusion module designed based on a cross-attention mechanism. ViT-FuseNet has two distinct advantages: i) it incorporates roadside LiDAR point clouds as additional inputs to enhance the 3D object detection capability of the vehicle; and ii) for the effective fusion of data from two different modal sensors, we employ a cross-attention mechanism for feature fusion, rather than directly merging camera features with point clouds at the raw data level. Extensive experiments are conducted using the DAIR-V2X Dataset to demonstrate the effectiveness of the proposed method. Compared with advanced cooperative perception methods, our method achieves a 6.17% improvement in 3D-mAP (IoU=0.5) and a 8.72% improvement in 3D-mAP (IoU=0.7). Moreover, the framework achieves the highest 3D-mAP (IoU=0.5) in all three object categories of benchmarks for single-vehicle perception.

This paper investigates efficient transmission design in a class of multi-user cooperation networks, in which multiple information sources intend to distribute their messages to a sufficient proportion of ambient destinations with the assistance of multiple relays. We apply a relaying scheme that combines non-orthogonal transmission with network coding techniques to balance channel consumption and inter-user interference. The sources and relays are divided into clusters, terminals within each of which are allowed to non-orthogonally access the same channel. A class of finite-field network codes are adopted in the relays. We provide the methods to derive the system transmission error probability and diversity-multiplexing tradeoff (DMT), over Nakagami-m fading channels. Through error probability and finite-SNR DMT analysis for certain clustering strategies, and infinite-SNR DMT analysis for general situations, we show that the considered relaying scheme can notably improve system performance over the conventional approach that demands only orthogonal transmission in network-coded cooperation networks.

In a vehicular communication network, how to efficiently utilize the limited network resources to meet the stringent transmission quality of service (QoS) requirements in the dynamically changing environment is a challenging task. This paper studies a typical vehicle-to-vehicle (V2V) communication scenario, in which multiple source nodes desire to deliver different types of messages to their respective destinations. Considering the case that accurate instantaneous global channel state information (CSI) is unavailable, properly selecting access channel, transmission power, and data rate at each source to optimize the links' performance is an extremely involved mixed-integer stochastic optimization problem. We propose applying the distributed multi-agent deep reinforcement learning technique to solve the problem. Specifically, a distributed multi-agent parameterized deep Q-network (DMA-PDQN) algorithm is employed to explore optimal decision in the discrete-continuous hybrid action space. The idea of federated meta learning (FML) is also adopted to tackle the non-stationarity issue and improve the training efficiency of distributed agents. Using energy efficiency (EE) as the target transmission objective, through extensive simulation we show that our method can achieve excellent convergence and performance.

Data delivery in vehicular networks confronts unique challenges due to vehicle movement and stringent quality-of-service (QoS) demands. This paper investigates the efficient transmission design problem in a typical vehicle-to-vehicle (V2V) communication network, where multiple source-destination pairs with different types of message transmission requirements coexist and the network environment dynamically changes. A sequential transmission decision framework based on the multi-objective optimization (MOO) theory is proposed to maximize the performance of each link, while ensuring the QoS of different messages. We integrate the Lyapunov optimization theory with the weighted Tchebycheff method, as well as other mathematical optimization tools, to transform the original problem into a solvable MOO problem at each time slot. Transmission actions can be determined continuously according to the environment change and time-related design goals. The effectiveness and efficiency of the proposed method are demonstrated through extensive simulation experiments.

In this paper, we investigate the information transmission in a typical 5G device-to-device (D2D) communication application scenario, i.e., a content distribution system with a number of information sources intending to broadcast their messages to multiple destinations in the vicinity. Due to the dynamic nature of wireless signal propagation links, it is hard to guarantee a satisfactory performance by direct source-destination transmissions, especially when the system is operating in a reuse-mode. Relays can be introduced to the system to solve this problem. However, adopting the conventional repetition coding at relays inefficiently utilizes the available resources, for the considered multi-source scenario. Therefore, we investigate applying a class of finite-field network codes at the relays, when potentially three types of relays are deployed to assist the information distribution process. We proposed the algorithms to derive the system outage probability and analyze the trade-off between energy efficiency and spectral efficiency. Our analytical and numerical results clearly demonstrate the potential of exploiting network-coded cooperative communications in future 5G D2D systems.