Search Results (20)

Convolutional neural networks (CNNs) have achieved remarkable success in computer vision tasks, driving the rapid development of hardware accelerators. However, memory efficiency remains a key challenge, as conventional accelerators adopt layer-by-layer processing, leading to frequent external memory accesses (EMAs) of intermediate feature data, which increase energy consumption and latency. While layer fusion has been proposed to enhance inter-layer feature reuse, existing approaches typically rely on fixed data management tailored to specific architectures, introducing on-chip memory overhead and requiring trade-offs with EMAs. Moreover, prevalent residual connections further weaken fusion benefits due to diverse data reuse distances. To address these challenges, we propose layer-centric tile fusion, which integrates residual data loading with feature merging by leveraging receptive field relationships among feature tiles. A reuse distance-aware caching strategy is introduced to support flexible storage for various data types. We also develop a modeling framework to analyze the trade-off between on-chip memory usage and EMA-induced energy-delay product (EDP). Experimental results demonstrate that our method achieves 5.04–43.44% EDP reduction and 20.28–58.33% memory usage reduction compared to state-of-the-art designs on ResNet-18 and SRGAN. Full article

Efficient task offloading for delay-sensitive applications, such as autonomous driving, presents a significant challenge in multi-hop Vehicular Edge Computing (VEC) networks, primarily due to high vehicle mobility, dynamic network topologies, and complex end-to-end congestion problems. To address these issues, this paper proposes a graph attention-based reinforcement learning algorithm, named GAPO. The algorithm models the dynamic VEC network as an attributed graph and utilizes a graph neural network (GNN) to learn a network state representation that captures the global topological structure and node contextual information. Building on this foundation, an attention-based Actor–Critic framework makes joint offloading decisions by intelligently selecting the optimal destination and collaboratively determining the ratios for offloading and resource allocation. A multi-objective reward function, designed to minimize task latency and to alleviate link congestion, guides the entire learning process. Comprehensive simulation experiments and ablation studies show that, compared to traditional heuristic algorithms and standard deep reinforcement learning methods, GAPO significantly reduces average task completion latency and substantially decreases backbone link congestion. In conclusion, by deeply integrating the state-aware capabilities of GNNs with the decision-making abilities of DRL, GAPO provides an efficient, adaptive, and congestion-aware solution to the resource management problems in dynamic VEC environments. Full article

As autonomous driving technologies advance, the need for human-in-the-loop systems becomes increasingly critical to ensure safety, adaptability, and public confidence. This paper presents the design and evaluation of a context-aware immersive teleoperation interface that integrates real-time simulation, virtual reality, and multimodal feedback to support remote interventions in emergency scenarios. Built on a modular ROS2 architecture, the system allows seamless transition between simulated and physical platforms, enabling safe and reproducible testing. The experimental results show a high task success rate and user satisfaction, highlighting the importance of intuitive controls, gesture recognition accuracy, and low-latency feedback. Our findings contribute to the understanding of human-robot interaction (HRI) in immersive teleoperation contexts and provide insights into the role of multisensory feedback and control modalities in building trust and situational awareness for remote operators. Ultimately, this approach is intended to support the broader acceptability of autonomous driving technologies by enhancing human supervision, control, and confidence. Full article

Autonomous vehicles are increasingly being adopted, with manufacturers competing to enhance automation capabilities. While full automation eliminates human input, lower levels still require driver intervention under specific conditions. This study presents the design and development of a prototype vehicle featuring both low- and high-level control systems, integrated with a 5G-based teleoperation interface that enables seamless switching between autonomous and remote-control modes. The system includes a malfunction surveillance unit that monitors communication latency and obstacle conditions, triggering a hardware-based emergency braking mechanism when safety thresholds are exceeded. Field experiments conducted over four test phases around Chulalongkorn University demonstrated stable performance under both driving modes. Mean lateral deviations ranged from 0.19 m to 0.33 m, with maximum deviations up to 0.88 m. Average end-to-end latency was 109.7 ms, with worst-case spikes of 316.6 ms. The emergency fallback system successfully identified all predefined fault conditions and responded with timely braking. Latency-aware stopping analysis showed an increase in braking distance from 1.42 m to 2.37 m at 3 m/s. In scenarios with extreme latency (>500 ms), the system required operator steering input or fallback to autonomous mode to avoid obstacles. These results confirm the platform’s effectiveness in real-world teleoperation over public 5G networks and its potential scalability for broader deployment. Full article

Cooperative Intelligent Transportation Systems (C-ITSs) are emerging as transformative technologies, paving the way for safe and fully automated driving solutions. As the demand for autonomous vehicles accelerates, the development of advanced Radio Access Technologies capable of delivering reliable, low-latency vehicular communications has become paramount. Standardized approaches for Vehicular-to-Everything (V2X) communication often fall short in addressing the dynamic and diverse requirements of multi-service, multi-priority systems. Conventional vehicular networks employ static parameters such as Access Category (AC) in IEEE 802.11p-based ITS-G5 and Resource Reservation Interval (RRI) in C-V2X PC5 for prioritizing different V2X services. This static parameter assignment performs unsatisfactorily in dynamic and diverse requirements. To bridge this gap, we propose intelligent Multi-Attribute Decision-Making algorithms for adaptive AC selection in ITS-G5 and RRI adjustment in C-V2X PC5, tailored to the varying priorities of active V2X services. These adaptations are integrated with a priority-aware rate-control mechanism to enhance congestion management. Through extensive simulations conducted using NS3, our proposed strategies demonstrate superior performance compared to standardized methods, achieving improvements in one-way end-to-end latency, Packet Reception Ratio (PRR) and overall communication reliability. Full article

With the rapid development of intelligent driving technology, in-vehicle bus networks face increasingly stringent requirements for real-time performance and data transmission. Traditional bus network technologies such as LIN, CAN, and FlexRay are showing significant limitations in terms of bandwidth and response speed. In-Vehicle Ethernet, with its advantages of high bandwidth, low latency, and high reliability, has become the core technology for next-generation in-vehicle communication networks. This study focuses on bandwidth waste caused by guard bands and the limitations of Frame Pre-Emption in fully utilizing available bandwidth in In-Vehicle Ethernet. It aims to optimize TSN scheduling mechanisms by enhancing scheduling flexibility and bandwidth utilization, rather than modeling system-level vehicle functions. Based on the Time-Sensitive Networking (TSN) protocol, this paper proposes an innovative Adaptive Frame Segmentation (AFS) algorithm. The AFS algorithm enhances the performance of In-Vehicle Ethernet message transmission through flexible frame segmentation and efficient message scheduling. Experimental results indicate that the AFS algorithm achieves an average local bandwidth utilization of 94.16%, improving by 4.35%, 5.65%, and 30.48% over Frame Pre-Emption, Packet-Size Aware Scheduling (PAS), and Improved Qbv algorithms, respectively. The AFS algorithm demonstrates stability and efficiency in complex network traffic scenarios, reducing bandwidth waste and improving In-Vehicle Ethernet’s real-time performance and responsiveness. This study provides critical technical support for efficient communication in intelligent connected vehicles, further advancing the development and application of In-Vehicle Ethernet technology. Full article

Mechanically scanning LiDAR imaging sensors are abundantly used in applications ranging from basic safety assistance to high-level automated driving, offering excellent spatial resolution and full surround-view coverage in most scenarios. However, their complex optomechanical structure introduces limitations, namely limited mounting options and blind zones, especially in elongated vehicles. To mitigate these challenges, we propose a distributed Time-of-Flight (ToF) sensor system with a flexible hardware–software architecture designed for multi-sensor synchronous triggering and fusion. We formalize the sensor triggering, interference mitigation scheme, data aggregation and fusion procedures and highlight challenges in achieving accurate global registration with current state-of-the-art methods. The resulting surround view visual information is then applied to Spiking Neural Network (SNN)-based object detection and probabilistic occupancy grid mapping (OGM) for enhanced environmental awareness. The proposed system is demonstrated on a test vehicle, achieving coverage of blind zones in a range of 0.5–6 m with a scalable and reconfigurable sensor mounting setup. Using seven ToF sensors, we can achieve a 10 Hz synchronized frame rate, with a 360° point cloud registration and fusion latency below 40 ms. We collected real-world driving data to evaluate the system, achieving 65% mean Average Precision (mAP) in object detection with our SNN. Overall, this work presents a replacement or addition to LiDAR in future high-level automation tasks, offering improved coverage and system integration. Full article

Integrated Sensing, Communication, and Computation (ISCC) has become a key technology driving the development of the Internet of Vehicles (IoV) by enabling real-time environmental sensing, low-latency communication, and collaborative computing. However, the increasing sensing data within the IoV leads to demands of fast data transmission in the context of limited communication resources. To address this issue, we propose a Collaborative Sensing-Aware Task Offloading (CSTO) mechanism for ISCC to reduce the sensing tasks transmission delay. We formulate a joint task offloading and communication resource allocation optimization problem to minimize the total processing delay of all vehicular sensing tasks. To solve this mixed-integer nonlinear programming (MINLP) problem, we design a two-stage iterative optimization algorithm that decomposes the original optimization problem into a task offloading subproblem and a resource allocation subproblem, which are solved iteratively. In the first stage, a Deep Reinforcement Learning algorithm is used to determine task offloading decisions based on the initial setting. In the second stage, a convex optimization algorithm is employed to allocate communication bandwidth according to the current task offloading decisions. We conduct simulation experiments by varying different crucial parameters, and the results demonstrate the superiority of our scheme over other benchmark schemes. Full article

The Internet of Things (IoT) has vast potential to drive connectivity and automation across various sectors, yet energy inefficiency remains a critical barrier to achieving sustainable, high-performing networks. This study aims to identify and address the primary causes of energy wastage in IoT systems, proposing a framework to optimise energy consumption and improve overall system performance. A comprehensive literature review was conducted, focusing on studies from 2010 onwards across major databases, resulting in the identification of eleven key factors driving energy inefficiency: offloading, scheduling, latency, changing topology, load balancing, node deployment, resource management, congestion, clustering, routing, and limited bandwidth. The impact of each factor on energy usage was analysed, leading to a proposed framework that incorporates optimised communication protocols (such as CoAP and MQTT), adaptive fuzzy logic systems, and bio-inspired algorithms to streamline resource management and enhance network stability. This framework presents actionable strategies to improve IoT energy efficiency, extend device lifespan, and reduce operational costs. By addressing these energy inefficiency challenges, this study provides a path forward for more sustainable IoT systems, emphasising the need for continued research into experimental validations, context-aware solutions, and AI-driven energy management to ensure scalable and resilient IoT deployment. Full article

The Internet of Things (IoT) presents immense opportunities for driving Industry 4.0 forward. However, in scenarios involving networked control automation, ensuring high reliability and predictable latency is vital for timely responses. To meet these demands, the contemporary wireless protocol time-slotted channel hopping (TSCH), also referred to as IEEE 802.15.4-2015, relies on precise transmission schedules to prevent collisions and achieve consistent end-to-end traffic flow. In the realm of diverse IoT applications, this study introduces a new traffic-aware autonomous multi-objective scheduling function called OPTIMAOrchestra. This function integrates slotframe and channel management, adapts to varying network sizes, supports mobility, and reduces collision risks. The effectiveness of two versions of OPTIMAOrchestra is extensively evaluated through multi-run experiments, each spanning up to 3600 s. It involves networks ranging from small-scale setups to large-scale deployments with 111 nodes. Homogeneous and heterogeneous network topologies are considered in static and mobile environments, where the nodes within a network send packets to the server with the same and different application packet intervals. The results demonstrate that OPTIMAOrchestra_ch4 achieves a current consumption of 0.72 mA while maintaining 100% reliability and 0.86 mA with a 100% packet delivery ratio in static networks. Both proposed Orchestra variants in mobile networks achieve 100% reliability, with current consumption recorded at 6.36 mA. Minimum latencies of 0.073 and 0.02 s are observed in static and mobile environments, respectively. On average, a collision rate of 5% is recorded for TSCH and RPL communication, with a minimum of 0% collision rate observed in the TSCH broadcast in mobile networks. Overall, the proposed OPTIMAOrchestra scheduler outperforms existing schedulers regarding network efficiency, time, and usability, significantly improving reliability while maintaining a balanced latency–energy trade-off. Full article

The platooning of cars and trucks is a pertinent approach for autonomous driving due to the effective utilization of roadways. The decreased gas consumption levels are an added merit owing to sustainability. Conventional platooning depended on Dedicated Short-Range Communication (DSRC)-based vehicle-to-vehicle communications. The computations were executed by the platoon members with their constrained capabilities. The advent of 5G has favored Intelligent Transportation Systems (ITS) to adopt Multi-access Edge Computing (MEC) in platooning paradigms by offloading the computational tasks to the edge server. In this research, vital parameters in vehicular platooning systems, viz. latency-sensitive radio resource management schemes, and Age of Information (AoI) are investigated. In addition, the delivery rates of Cooperative Awareness Messages (CAM) that ensure expeditious reception of safety-critical messages at the roadside units (RSU) are also examined. However, for latency-sensitive applications like vehicular networks, it is essential to address multiple and correlated objectives. To solve such objectives effectively and simultaneously, the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) framework necessitates a better and more sophisticated model to enhance its ability. In this paper, a novel Cascaded MADDPG framework, CMADDPG, is proposed to train cascaded target critics, which aims at achieving expected rewards through the collaborative conduct of agents. The estimation bias phenomenon, which hinders a system’s overall performance, is vividly circumvented in this cascaded algorithm. Eventually, experimental analysis also demonstrates the potential of the proposed algorithm by evaluating the convergence factor, which stabilizes quickly with minimum distortions, and reliable CAM message dissemination with $99\%$ probability. The average AoI quantity is maintained within the 5–10 ms range, guaranteeing better QoS. This technique has proven its robustness in decentralized resource allocation against channel uncertainties caused by higher mobility in the environment. Most importantly, the performance of the proposed algorithm remains unaffected by increasing platoon size and leading channel uncertainties. Full article

The expansion of mobile connectivity with the arrival of 6G paves the way for the new Internet of Verticals (6G-IoV), benefiting autonomous driving. This article highlights the importance of vehicle-to-everything (V2X) and vehicle-to-vehicle (V2V) communication in improving road safety. Current technologies such as IEEE 802.11p and LTE-V2X are being improved, while new radio access technologies promise more reliable, lower-latency communications. Moreover, 3GPP is developing NR-V2X to improve the performance of communications between vehicles, while IEEE proposes the 802.11bd protocol, aiming for the greater interoperability and detection of transmissions between vehicles. Both new protocols are being developed and improved to make autonomous driving more efficient. This study analyzes and compares the performance of the protocols mentioned, namely 802.11p, 802.11bd, LTE-V2X, and NR-V2X. The contribution of this study is to identify the most suitable protocol that meets the requirements of V2V communications in autonomous driving. The relevance of V2V communication has driven intense research in the scientific community. Among the various applications of V2V communication are Cooperative Awareness, V2V Unicast Exchange, and V2V Decentralized Environmental Notification, among others. To this end, the performance of the Link Layer of these protocols is evaluated and compared. Based on the analysis of the results, it can be concluded that NR-V2X outperforms IEEE 802.11bd in terms of transmission latency (L) and data rate (DR). In terms of the packet error rate (PER), it is shown that both LTE-V2X and NR-V2X exhibit a lower PER compared to IEEE protocols, especially as the distance between the vehicles increases. This advantage becomes even more significant in scenarios with greater congestion and network interference. Full article

Cooperative communications advancements in Vehicular-to-Everything (V2X) are bolstering the autonomous driving paradigm. V2X nodes are connected through communication technology, such as a short-range communication mode (Dedicated Short Range Communication (DSRC) and Cellular-V2X) or a long-range communication mode (Uu). Conventional vehicular networks employ static wireless vehicular communication technology without considering the traffic load on any individual V2X communication technology and the traffic dynamics in the vicinity of the V2X node, and are hence inefficient. In this study, we investigate hybrid V2X communication and propose an autonomous and intelligent technology selection algorithm using a decision tree. The algorithm uses the information from the received Cooperative Intelligent Transport Systems (C-ITS) Cooperative Awareness Messages (CAMs) to collect statistics such as inter vehicular distance, one-way end-to-end latency and CAM density. These statistics are then used as input for the decision tree for selecting the appropriate technology (DSRC, C-V2X PC5 or 5G) for the subsequent scheduled C-ITS message transmission. The assessment of the intelligent hybrid V2X algorithm’s performance in our V2X test setup demonstrates enhancements in one-way end-to-end latency, reliability, and packet delivery rate when contrasted with the conventional utilization of static technology. Full article

Although NAND (Not And) flash-based Solid-State Drive (SSD) has recently demonstrated a significant performance advantage against hard disk, it still suffers from non-negligible performance under-utilization issues as the access conflict often occurs during servicing IO requests due to the share mechanism (e.g., several chips share one channel bus, several planes share one data register inside the die). Many research works have been devoted to minimizing access conflict by redesigning IO scheduling, cache replacement, and so on. These works have achieved reasonable results; however, the potential data similarity characterization is not utilized fully in prior works to alleviate access conflict. The basic idea is that, as data duplication is common in many workloads where data with the same content from different requests could be distributed to the address with minimized access conflict (i.e., the address does not share the same channel or chip), the logic address is mapped to more than one physical address. Therefore, the data can be read out from candidate pages when the channel or chip of its original address is busy. Motivated by this idea, we propose Data Similarity aware Flash Translation Layer (DS-FTL), which mainly includes a content-aware page allocation scheme and a multi-path read scheme. The DS-FTL enables maximization of the channel-level and chip-level parallelism and avoids the read stall induced by bus-shared mechanisms. We also conducted a series of experiments on SSDsim, with the subsequent results depicting the effectiveness of our scheme. Compared with the state-of-art, our scheme reduces read latency by 35.3% on average in our workloads. Full article

When deciding whether to cross the street or not, pedestrians take into consideration information provided by both vehicle kinematics and the driver of an approaching vehicle. It will not be long, however, before drivers of autonomous vehicles (AVs) will be unable to communicate their intention to pedestrians, as they will be engaged in activities unrelated to driving. External human–machine interfaces (eHMIs) have been developed to fill the communication gap that will result by offering information to pedestrians about the situational awareness and intention of an AV. Several anthropomorphic eHMI concepts have employed facial expressions to communicate vehicle intention. The aim of the present study was to evaluate the efficiency of emotional (smile; angry expression) and conversational (nod; head shake) facial expressions in communicating vehicle intention (yielding; non-yielding). Participants completed a crossing intention task where they were tasked with deciding appropriately whether to cross the street or not. Emotional expressions communicated vehicle intention more efficiently than conversational expressions, as evidenced by the lower latency in the emotional expression condition compared to the conversational expression condition. The implications of our findings for the development of anthropomorphic eHMIs that employ facial expressions to communicate vehicle intention are discussed. Full article