TO: Prof. Ellis
FROM: DeAndre Badresingh
DATE: Oct. 6, 2021
SUBJECT: 500-Word Summary of Article About Malware Detection In Self-Driving Cars
Since transportation becomes more intelligent, it leaves it more vulnerable to cyber-attacks. There has been many times where users have lost control of their vehicle due to someone attacking their system. This is typically caused by various forms of malicious software. Malicious software replicates an already authorized software for self-driving vehicles. Methods and experiments have been put In place to analyze detection of compromised self- driving cars. Vehicle to vehicle protection is important because it allows external connections to not only provide comfort for the driver but also update the security. Security technology is analyzed and scans the security of the car for intrusion detection.
The main method for hackers getting access to information is through the use of malicious code which allows them to gain or deny access to a user’s system. To combat this, Machine learning algorithm using the software called Adware and General Malware (AW&GM) is used to differentiate normal code from malicious code used by hackers.
Attempted breaches come in many forms which includes malicious messages, denial of service, or even adware. A method which involves reconfiguring electronic control units uses a control module known as mitigation manager that scans for cyber-attacks. Another method for controlling these types of attacks involves an algorithm that scans for unusual patterns in within the vehicles network. Furthermore, another concept in mind was the use of cloud defense framework which allows only one gateway to monitor all traffic going into and out of the network.
Since self-driving vehicles are usually connected to public networks security is key to protecting them due to higher chances of having the operating system compromised. On a most recent machine learning algorithm study, intrusion detection was installed into the vehicles which allows the unit to actively see real-time changes in behavioral rhythm. With this new software, the algorithm can determine the intrusion more accurately by learning, verifying, and evaluating messaging patterns.
In the event of unusually high network traffic, intrusion detection relies on scaling. Scaling prevents under or overflow of data when undergoing experiments. When the environment is right, the multiple rounds of test begins. More tests are required during the experimental phase because they may come back as false positives.
To conclude, IDS go through the three phases of data preprocessing, modeling, and detecting. Simulated results are compared to proposed algorithms. Benign code, adware, and general malware are known as classification scenarios.
Using random forest, also known as RF, has been proven to have a higher predication accuracy. It has been concluded that using an algorithm with short learning time can use used to prove the mode accurate results. Receiver operating characterizes are also used to calculate the results from the tests. Each use of these methods revealed to have a different success result. Since transportation is ever-changing, security has to keep up to protect users.
Reference:
[1] S. Park and J.-Y. Choi, “Malware detection in self-driving vehicles using machine learning algorithms,” Journal of Advanced Transportation, vol. 2020, pp. 1–9, 2020.
