Scholarlink

The rapid advancement of autonomous vehicles has revolutionized modern transportation, integrating intelligent systems that rely heavily on the Internet of Things (IoT). These vehicles are equipped with numerous IoT-enabled sensors and communication modules to ensure real-time decision-making, route planning, obstacle avoidance, and vehicle-to-everything (V2X) communication. The cybersecurity-related balance of data communication methods utilized in autonomous vehicles may increase more exposure in the data communication procedure among these vehicles and IoT devices. Securing autonomous vehicle systems (AVS) is dominant to ensure the safety and trustworthiness of autonomous transport. Cybersecurity recognition includes the execution of innovative threat detection devices to recognize and moderate potential cyber-attacks. Applying a combination of anomaly recognition, machine learning (ML) and encryption methods, these systems can able to monitor network traffic, detect strange patterns, and quickly respond to any malicious actions. The exclusive tasks modeled by autonomous vehicles like the requirement for real-world decision-making and the huge extent of data produced, need robust and adaptive cybersecurity methods. To resolve these issues, this manuscript proposes a novel Cybersecurity Detection using a Hybrid Deep Learning Approach for Self-Driving Vehicle Networks, named CDHDL-SDVN technique in IoT. The CDHDL-SDVN technique aims to recognize cyberattacks in self-driving vehicle networks using a hyperparameter-tuned ensemble deep learning (DL) model. At the initial stage, the CDHDL-SDVN technique utilizes a linear scalar normalization approach for data pre-processing. Besides, an artificial gorilla troops optimizer (GTO)-based feature selection process is performed to reduce the high dimensionality problem. Moreover, the cyberattack detection is performed by employing an ensemble of long short-term memory (LSTM), deep extreme learning machines (DELM), and gated recurrent units (GRUs) models. At last, the hyperparameter tuning of DL models is accomplished by using an artificial ecosystem-based optimization (AEO) approach. A series of tests were accomplished to test the performance of the CDHDL-SDVN technique. The experimentation validation of the CDHDL-SDVN technique attains a superior accuracy value of 98.61% over existing techniques in enhancing cybersecurity in self-driving vehicle networks.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Internet of Things-Enabled Cyber Threat Detection in Self-driving Vehicle Networks Using a Hybrid Deep Learning-Based Security Model

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Tareq M. Alkhaldi

摘要

The rapid advancement of autonomous vehicles has revolutionized modern transportation, integrating intelligent systems that rely heavily on the Internet of Things (IoT). These vehicles are equipped with numerous IoT-enabled sensors and communication modules to ensure real-time decision-making, route planning, obstacle avoidance, and vehicle-to-everything (V2X) communication. The cybersecurity-related balance of data communication methods utilized in autonomous vehicles may increase more exposure in the data communication procedure among these vehicles and IoT devices. Securing autonomous vehicle systems (AVS) is dominant to ensure the safety and trustworthiness of autonomous transport. Cybersecurity recognition includes the execution of innovative threat detection devices to recognize and moderate potential cyber-attacks. Applying a combination of anomaly recognition, machine learning (ML) and encryption methods, these systems can able to monitor network traffic, detect strange patterns, and quickly respond to any malicious actions. The exclusive tasks modeled by autonomous vehicles like the requirement for real-world decision-making and the huge extent of data produced, need robust and adaptive cybersecurity methods. To resolve these issues, this manuscript proposes a novel Cybersecurity Detection using a Hybrid Deep Learning Approach for Self-Driving Vehicle Networks, named CDHDL-SDVN technique in IoT. The CDHDL-SDVN technique aims to recognize cyberattacks in self-driving vehicle networks using a hyperparameter-tuned ensemble deep learning (DL) model. At the initial stage, the CDHDL-SDVN technique utilizes a linear scalar normalization approach for data pre-processing. Besides, an artificial gorilla troops optimizer (GTO)-based feature selection process is performed to reduce the high dimensionality problem. Moreover, the cyberattack detection is performed by employing an ensemble of long short-term memory (LSTM), deep extreme learning machines (DELM), and gated recurrent units (GRUs) models. At last, the hyperparameter tuning of DL models is accomplished by using an artificial ecosystem-based optimization (AEO) approach. A series of tests were accomplished to test the performance of the CDHDL-SDVN technique. The experimentation validation of the CDHDL-SDVN technique attains a superior accuracy value of 98.61% over existing techniques in enhancing cybersecurity in self-driving vehicle networks.