الفهرس 
Chapter 2 : Background and literature surveyOnly 14 pages are availabe for public view
 |  | from | 71 |  |  |
| | | from | 71 | | |
Abstract
Abstract
Vehicles currently are computerised, connected devices on the roads. The In-vehicle attacks are safety critical and life threatening, as intrusion does not affect data only like the other connected devices, it may affect people’s lives. Currently In Vehicle Intrusion detection systems (IDS) goes beyond just research, now it is a part of automotive Software[1].
Prevalent vehicles are connected to external world through different channels such as On-board diagnostics port (OBD2) which is considered as an interface between the tester hardware and the vehicle Electronic control unit (ECU), which used to control different functions of the vehicle ECUs such as flashing firmware or read/clear diagnostics data, This port is considered as big security hole for the in vehicle network as it has direct access to the vehicle Controller area network (CAN) bus, flashing over the air is another communication channel between the vehicle and the external world, Internet connection also is another communication channel as large number of modern vehicles have Infotainment systems connected to the internet. These communication channels may be used by adversaries to make malicious attacks to the vehicle networks, these attacks can be fatal if the attacker successfully controls the steering wheel or the braking system.
Different intrusion detection systems proposed to detect and protect the vehicles from these malicious attacks. We listed the recent in-vehicle intrusion detection systems and mentioned the advantages and disadvantages for each category.
We also proposed two implementation approaches, the first approach using traditional computing techniques for intrusion detection and the other is based on attention deep learning technique this technique gives a recall and F1 scores closes 99.9%, these results are outperforming the other deep learning-based techniques such as LSTM and ANN.