To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
A New Combination of Machine Learning Algorithms using Stacking Approach for Misbehavior Detection in VANETs |
|
Author |
Abhilash Sonker and Dr. R K Gupta" |
| Citation |
Vol. 20 No. 10 pp. 94-100 |
|
Abstract |
Road safety, traffic efficiency and passenger comfort are main reasons for the emergence of Vehicular Adhoc Networks (VANETs). The misbehavior in the nodes can be detected with its communication to other nodes. The performance of VANET applications depend on messages and information shared by vehicles. When a message is sent from one node to another node it has some features. With the study of these features it can be found that the message sent is malicious or not. The detection of malicious activity is hence an important component. In this paper, a new combination of machine learning algorithms using stacking approach is built to find the misbehavior in the message log sent by a node in VANETs. Correspondingly, it will be detected that the message sent from the node is malicious or not. A new combination is designed with Random Forest (bagging) and Xgboost (boosting) using stacking to get more accuracy in multiclass classification of attacks. With this new combination of algorithms using stacking 98.44% of accuracy is achieved. This accuracy is evaluated on the test data. For this work, VeReMi dataset (public dataset for the malicious node detection) is used. |
|
Keywords |
Misbehavior Detection; Machine Learning; Stacking Algorithm; Vehicular Adhoc Networks |
|
URL |
|
|
Title |
A New Combination of Machine Learning Algorithms using Stacking Approach for Misbehavior Detection in VANETs |
|
Author |
Abhilash Sonker and Dr. R K Gupta |
| Citation |
Vol. 20 No. 10 pp. 94-100 |
|
Abstract |
Road safety, traffic efficiency and passenger comfort are main reasons for the emergence of Vehicular Adhoc Networks (VANETs). The misbehavior in the nodes can be detected with its communication to other nodes. The performance of VANET applications depend on messages and information shared by vehicles. When a message is sent from one node to another node it has some features. With the study of these features it can be found that the message sent is malicious or not. The detection of malicious activity is hence an important component. In this paper, a new combination of machine learning algorithms using stacking approach is built to find the misbehavior in the message log sent by a node in VANETs. Correspondingly, it will be detected that the message sent from the node is malicious or not. A new combination is designed with Random Forest (bagging) and Xgboost (boosting) using stacking to get more accuracy in multiclass classification of attacks. With this new combination of algorithms using stacking 98.44% of accuracy is achieved. This accuracy is evaluated on the test data. For this work, VeReMi dataset (public dataset for the malicious node detection) is used. |
|
Keywords |
Misbehavior Detection; Machine Learning; Stacking Algorithm; Vehicular Adhoc Networks |
|
URL |
