Authors :
Ladan, Nanbal Jibba; Katniyon, Henry David; Pam, Bulus Dung; Ramson, Emmanuel Nannim; Datti Useni Emmanuel; Mullah Sallau Nanlir
Volume/Issue :
Volume 8 - 2023, Issue 2 - February
Google Scholar :
https://bit.ly/3IIfn9N
Scribd :
https://bit.ly/41yl5pf
DOI :
https://doi.org/10.5281/zenodo.7688801
Abstract :
The increased popularity being witnessed in
the Internet of Things (IoT) domain has brought with it
challenges in the area of security. From all indications,
this growth we are witnessing will be of exponential
proportions in the nearest future. The need to tackle
security challenges is of utmost importance. This study
was embarked upon to do exactly that. We were able to
gain access to Bot-IoT dataset which was suitable since it
was created specifically for IoT. A Deep Neural Network
(DNN) was deployed and used to train and validate our
dataset to predict and categorize the five types of botnet
attacks present in the dataset. DNN was able to do that
with an accuracy rate of 97%. Afterwards, a peer
reviewed journal article which had used other Machine
Learning (ML) models was selected and our results were
compared. After the comparison, it was observed that
RNN and LSTM had a slightly higher accuracy of 99%
each but our model had a higher accuracy rate than
SVM which stood at 88%.
Keywords :
Internet of Things, Deep Learning, Machine Learning, Botnet.
The increased popularity being witnessed in
the Internet of Things (IoT) domain has brought with it
challenges in the area of security. From all indications,
this growth we are witnessing will be of exponential
proportions in the nearest future. The need to tackle
security challenges is of utmost importance. This study
was embarked upon to do exactly that. We were able to
gain access to Bot-IoT dataset which was suitable since it
was created specifically for IoT. A Deep Neural Network
(DNN) was deployed and used to train and validate our
dataset to predict and categorize the five types of botnet
attacks present in the dataset. DNN was able to do that
with an accuracy rate of 97%. Afterwards, a peer
reviewed journal article which had used other Machine
Learning (ML) models was selected and our results were
compared. After the comparison, it was observed that
RNN and LSTM had a slightly higher accuracy of 99%
each but our model had a higher accuracy rate than
SVM which stood at 88%.
Keywords :
Internet of Things, Deep Learning, Machine Learning, Botnet.