Autonomous Vehicles: Sophisticated Attacks, Safety Issues, Challenges, Open Topics, Blockchain, and Future Directions

Journal of Cybersecurity and Privacy, Journal Year: 2023, Volume and Issue: 3(3), P. 493 - 543

Published: Aug. 5, 2023

Autonomous vehicles (AVs), defined as capable of navigation and decision-making independent human intervention, represent a revolutionary advancement in transportation technology. These operate by synthesizing an array sophisticated technologies, including sensors, cameras, GPS, radar, light imaging detection ranging (LiDAR), advanced computing systems. components work concert to accurately perceive the vehicle’s environment, ensuring capacity make optimal decisions real-time. At heart AV functionality lies ability facilitate intercommunication between with critical road infrastructure—a characteristic that, while central their efficacy, also renders them susceptible cyber threats. The potential infiltration these communication channels poses severe threat, enabling possibility personal information theft or introduction malicious software that could compromise vehicle safety. This paper offers comprehensive exploration current state technology, particularly examining intersection autonomous emotional intelligence. We delve into extensive analysis recent research on safety lapses security vulnerabilities vehicles, placing specific emphasis different types attacks which they are susceptible. further explore various solutions have been proposed implemented address discussion not only provides overview existing challenges but presents pathway toward future directions. includes advancements field, continued refinement measures, development more robust, resilient mechanisms. Ultimately, this seeks contribute deeper understanding landscape fostering discourse intricate balance technological rapidly evolving field.

Language: Английский

---

VeSoNet: Traffic-Aware Content Caching for Vehicular Social Networks Using Deep Reinforcement Learning

IEEE Transactions on Intelligent Transportation Systems, Journal Year: 2023, Volume and Issue: 24(8), P. 8638 - 8649

Published: March 14, 2023

Vehicular social networking is an emerging application of the Internet Vehicles (IoV) which aims to achieve seamless integration vehicular networks and networks. However, unique characteristics networks, such as high mobility frequent communication interruptions, make content delivery end-users under strict delay constraints extremely challenging. In this paper, we propose a social-aware edge computing architecture that solves problem by using some vehicles in network servers can store stream popular close-by end-users. The proposed includes three main components: 1) graph pruning search algorithm computes assigns shortest path with most relevant providers. 2) traffic-aware recommendation scheme recommends according its context. This uses embeddings are represented set low-dimension vectors (vehicle2vec) information about previously consumed content. Finally, deep reinforcement learning (DRL) method optimise provider vehicle distribution across network. results obtained from real-world traffic simulation show effectiveness robustness system when compared state-of-the-art baselines.

Language: Английский

---

Multi-Agent Reinforcement Learning for Traffic Flow Management of Autonomous Vehicles

Sensors, Journal Year: 2023, Volume and Issue: 23(5), P. 2373 - 2373

Published: Feb. 21, 2023

Intelligent traffic management systems have become one of the main applications Transportation Systems (ITS). There is a growing interest in Reinforcement Learning (RL) based control methods ITS such as autonomous driving and solutions. Deep learning helps approximating substantially complex nonlinear functions from complicated data sets tackling issues. In this paper, we propose an approach on Multi-Agent (MARL) smart routing to improve flow vehicles road networks. We evaluate Advantage Actor-Critic (MA2C) Independent Actor-Critical (IA2C), recently suggested techniques with for signal optimization determine its potential. investigate framework offered by non-Markov decision processes, enabling more in-depth understanding algorithms. conduct critical analysis observe robustness effectiveness method. The method's efficacy reliability are demonstrated simulations using SUMO, software modeling tool simulations. used network that contains seven intersections. Our findings show MA2C, when trained pseudo-random vehicle flows, viable methodology outperforms competing techniques.

Language: Английский

---

The deceitful Connected and Autonomous Vehicle: Defining the concept, contextualising its dimensions and proposing mitigation policies

Transport Policy, Journal Year: 2022, Volume and Issue: 122, P. 1 - 10

Published: April 21, 2022

The Connected and Autonomous Vehicle (CAV) is an emerging mobility technology that may hold a paradigm-changing potential for the future of transport policy planning. Despite wealth likely benefits have made their eventual launch inescapable, CAVs also be source unprecedented disruption tomorrow's travel eco-systems because vulnerability to cyber-threats, hacking misinformation. manipulated by users, traffic controllers or third parties act in deceitful ways. This scene-setting work introduces CAV, vehicle operates manner towards routing control functionality 'selfish' malicious purposes contextualises its diverse expressions dimensions. It specifically offers systematic taxonomy eight distinctive behaviours namely: suppression/camouflage, overloading, mistake, substitution, target conditioning, repackaging capability signatures, amplification reinforcing impression. These as exemplified most common attack forms (i.e., starvation, denial-of-service, session hijacking, man-in-the-middle, poisoning, masquerading, flooding spoofing) are then benchmarked against five key dimensions referring time frame (short long duration), engagement (localised systemic), urban controller infrastructure (single multiple components), scale (low high), impact high). We suggest mitigation strategies protect CAV these dangers. span from purely technological measures machine-centric triad vehicles, communication, system including adversarial training, heuristic decision algorithms weighted voting mechanisms human factor focus on education, awareness enhancement, licensing legislation initiatives will enable users prevent, report activities.

Language: Английский

---

Non-divergent traffic management scheme using classification learning for smart transportation systems

Computers & Electrical Engineering, Journal Year: 2023, Volume and Issue: 106, P. 108581 - 108581

Published: Jan. 11, 2023

Language: Английский

---

Autonomous and Human-Driven Vehicles Interacting in a Roundabout: A Quantitative and Qualitative Evaluation

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 32693 - 32705

Published: Jan. 1, 2024

Optimizing traffic dynamics in an evolving transportation landscape is crucial, particularly scenarios where autonomous vehicles (AVs) with varying levels of autonomy coexist human-driven cars. While optimizing Reinforcement Learning (RL) policies for such becoming more and common, little has been said about realistic evaluations trained policies. This paper presents evaluation the effects AVs penetration among human drivers a roundabout scenario, considering both quantitative qualitative aspects. In particular, we learn policy to minimize jams (i.e., time cross scenario) pollution Milan, Italy. Through empirical analysis, demonstrate that presence can reduce levels. Furthermore, qualitatively evaluate learned using cutting-edge cockpit assess its performance near-real-world conditions. To gauge practicality acceptability policy, conduct participants simulator, focusing on range metrics like smoothness safety perception. general, our findings show benefit from dynamics. Also, study highlight scenario 80% perceived as safer than 20%. The same result obtained

Language: Английский

---

Optimizing traffic flow with Q-learning and genetic algorithm for congestion control

Evolutionary Intelligence, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 4, 2024

Language: Английский

---

Drone Navigation Using Region and Edge Exploitation-Based Deep CNN

IEEE Access, Journal Year: 2022, Volume and Issue: 10, P. 95441 - 95450

Published: Jan. 1, 2022

Drones are unmanned aerial vehicles (UAV) utilized for a broad range of functions, including delivery, surveillance, traffic monitoring, architecture and even War-field. confront significant obstacles while navigating independently in complex highly dynamic environments. Moreover, the targeted objects within environment have irregular morphology, occlusion, minor contrast variation with background. In this regard, novel deep Convolutional Neural Network(CNN) based data-driven strategy is proposed drone navigation environment. The Drone Split-Transform-and-Merge Region-and-Edge (Drone-STM-RENet) CNN comprised convolutional blocks where each block methodically implements region edge operations to preserve diverse set properties at multi-levels, especially congested block, systematic implementation average max-pooling can deal homogeneity properties. Additionally, these merged multi-level learn texture that efficiently discriminates target from background helps obstacle avoidance. Finally, Drone-STM-RENet generates steering angle collision probability input image control moving avoiding hindrances allowing UAV spot risky situations respond quickly, respectively. has been validated on two urban cars bicycles datasets: udacity collision-sequence, achieved considerable performance terms explained variance (0.99), recall (95.47%), accuracy (96.26%), F-score (91.95%). promising road datasets suggests model generalizable be deployed real-time autonomous drones real-world flights.

Language: Английский

---

Adaptive Traffic Control Using Cooperative Communication Through Visible Light

SN Computer Science, Journal Year: 2024, Volume and Issue: 5(1)

Published: Jan. 6, 2024

Abstract The study aims to create a Visible-Light Communication (VLC) system for secure vehicle management at intersections. This involves enabling communication between vehicles and infrastructure (V2V, V2I, I2V) using headlights, streetlights, traffic signals. Mobile optical receivers gather data, determine their location, read transmitted information through joint transmission. An intersection manager coordinates communicates with embedded Driver Agents. utilizes "mesh/cellular" hybrid network configuration encodes data into light signals emitted by transmitters. Optical sensors filtering properties enable reception decoding. demonstrates bidirectional communication, employing queue/request/response mechanisms relative pose concepts safe passage. A deep reinforcement learning model controls cycles, validated via simulation in Simulation of Urban Mobility simulator. Results show that this adaptive control effectively collects detailed ensures within the short-range mesh network.

Language: Английский

---

Implementation of Controlling the Traffic Light System Using RQL

Wireless Personal Communications, Journal Year: 2024, Volume and Issue: 135(1), P. 451 - 491

Published: March 1, 2024

Language: Английский

---