期刊
IEEE SENSORS JOURNAL
卷 21, 期 14, 页码 15859-15866出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JSEN.2020.3012046
关键词
Deep learning; software defined intelligent transportation systems (SD-ITS) cyber threats & attacks; data-driven intelligent transportation systems
资金
- European Commission, under the ASTRID project [786922]
- European Commission, under FutureTPM project [779391]
- H2020 Societal Challenges Programme [786922] Funding Source: H2020 Societal Challenges Programme
The study proposes a DL-driven attack detection framework using GPU to address the increasingly complex attacks on SD-ITS. Experimental results demonstrate that the technique performs well in terms of detection accuracy with low computational complexity.
Intelligent transportation systems have been envisioned to bring more intelligence and cooperative sensing to meet the imminent demands of overall improved autonomous transportation. However, dynamic era of modern applications and fixed architecture of legacy Internet needs flexible, innovative, adaptive, and programmable software defined intelligent transportation systems (SD-ITS). The centralized control intelligence of SD-ITS can be a potential primary target of the prevalent cyber threats and attacks that can simply throw the entire network into chaos. The authors propose a DL-driven multi-vector scalable attack detection framework leveraging graphical processing unit (GPU) empowered Bidirectional Long Short-Term Memory (BLSTM) to efficiently tackle exponentially growing diverse sophisticated attacks that primarily target the control unit of the SD-ITS. The proposed technique has been rigorously evaluated with current state-of-the-art publicly available Flow-based dataset (i.e., CICIDS2017) using standard performance metrics. Further, the proposed mechanism is compared with contemporary benchmarks (i.e., DL algorithms). Extensive experimental results exhibit out-performance of the proposed technique in term of detection accuracy with a trivial trade-off computational complexity. Finally, the study also employed 10-fold cross validation to explicitly show unbiased results.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据