Article Versions
Export Article
Cite this article
  • Normal Style
  • MLA Style
  • APA Style
  • Chicago Style
Research Article
Open Access Peer-reviewed

A Modern Analysis of Aging Machine Learning Based IoT Cybersecurity Methods

Sam Strecker, Rushit Dave , Nyle Siddiqui, Naeem Seliya
Journal of Computer Sciences and Applications. 2021, 9(1), 16-22. DOI: 10.12691/jcsa-9-1-2
Received September 04, 2021; Revised October 09, 2021; Accepted October 18, 2021

Abstract

Modern scientific advancements often contribute to the introduction and refinement of never-before-seen technologies. This can be quite the task for humans to maintain and monitor and as a result, our society has become reliant on machine learning to assist in this task. With new technology comes new methods and thus new ways to circumvent existing cyber security measures. This study examines the effectiveness of three distinct Internet of Things cyber security algorithms currently used in industry today for malware and intrusion detection: Random Forest (RF), Support-Vector Machine (SVM), and K-Nearest Neighbor (KNN). Each algorithm was trained and tested on the Aposemat IoT-23 dataset which was published in January 2020 with the earliest of captures from 2018 and latest from 2019. The RF, SVM, and KNN reached peak accuracies of 92.96%, 86.23%, and 91.48%, respectively, in intrusion detection and 92.27%, 83.52%, and 89.80% in malware detection. It was found all three algorithms are capable of being effectively utilized for the current landscape of IoT cyber security in 2021.

1. Introduction

The number of Internet of Things (IoT) devices connected to the Internet has exceeded fifty billion devices since 2020 1, 2. This results in a dire need to continually advance the cyber security field as stated in 3. Previously vetted and approved security techniques need to be continually re-tested with new datasets. New and emerging technologies may deviate from past implementations and introduce new outlets for malicious users to exploit. For example, Google’s Home Mini uses a hard-coded DNS to try and deter users from DNS based adblocking, but the result is a new security threat. The device overrides the DNS set on the local network, and in this case uses Google’s DNS (8.8.8.8/8.8.4.4). Provided more information, a malicious user can now exploit the device with a current zero day named ‘Name:Wreck’. This is simply one example of thousands that cyber security analysts need to be aware of first before it is in the hands of a malicious user. This article will compare the effectiveness of three different machine learning based cyber security techniques on the Aposemat Iot-23 dataset published in January of 2020 4. Further, we produce an analysis of aging machine learning based IoT cybersecurity methods on newly published datasets that bring forth never-before seen IoT devices and their accompanying technologies/protocols, as seen in 5. We test RF, SVM, and K-Nearest Neighbor algorithms to be used for IoT cyber security in the present year of 2021. Section 2 will discuss previous application of IoT and machine learning in academic literature, Section 3 will discuss in detail the dataset and algorithms we use in this article, Sections 4 and 5 serve as an exhibition and analysis of our observed results, and Section 6 summarizes and concludes our work.

2. Background

The phrase “Internet of Things” has been around for over two decades. Created in 1999, 6 initially used the term to refer to the use of radio-frequency identification tags in an assembly line. Today the phrase IoT is more encompassing and includes any device that has the ability to collect data and communicate this data via the internet 7, 8, 9. Due to its usefulness and versatility, IoT devices became deeply integrated into the workings of society. From healthcare, to e-commerce, to one’s own living room, IoT devices are abundant in presence. For instance, the city of Padua, in Italy, utilizes IoT networks to monitor carbon monoxide levels, traffic flow, noise levels, streetlights, power usage, and much more as stated in 10. With the rise of technological prowess, so does the sophistication of malicious attacks on these technologies, thus making the need for stronger cyber security methods imperative 11. New methods to protect IoT devices from malicious users are continuously being developed, however, developing well-made cyber security methods for IoT devices comes with its own unique challenges. IoT devices are very resource limited, similar in device signature, prone to botnet attacks, and can be found in edge computing scenarios. This results in the necessity for unique cyber security that can accommodate these limitations. Researchers in 9 and 12 studied the effectiveness of (CoAP) - the constrained application protocol - which presented a cost-effective solution to protect data transmission within IoT networks in real-time. Through the optimization of this protocol, what would have normally been unachievable with current cyber security techniques was proven possible in 9 and 12 with accuracy rates reaching a peak of 97%. Overall, IoT cybersecurity could be greatly improved when compared to other areas like anti-fraud for online transactions 13. With the ever-increasing number of IoT devices on the internet, the occurrences of botnet attacks are growing proportionally to the volumetric traffic of each attack in accordance with 11. 14 exhibited the power of botnets when a peak Distributed Denial of Service (DDOS) attack with a bandwidth of 1.1Tb/s targeted the cloud service company Cloudflare, pushing it over the limit to what Cloudflare could mitigate resulting in large network outages. Current IoT security techniques rely on three main machine learning algorithms for device identification, intrusion attacks, and malware identification: SVMs, RFs, and KNNs. In essence, these three algorithms can be effectively utilized as watch dogs on IoT networks to prevent attacks in real-time. This study examines the effectiveness of RFs, SVMs, and KNNs for IoT cybersecurity against modern datasets.

According to 3, the use of supervised learning techniques like SVM, KNN, RF, Naïve Bayes, and artificial neural networks (ANN) are effective in identifying network traffic of IoT devices. More specifically, they have the capability to detect network intrusions as well as spoofing attacks. To identify Denial of Service (DoS) attacks, the use of multivariate correlation is needed. It extracts the geometrical correlations of network traffic features making the model 92% more accurate. Excluding deep learning, the researchers concluded that the RF performed most optimally for malware detection and KNN performed best for network intrusion. In 15, researchers set out to find a machine learning algorithm that had the ability to detect DDOS attacks, also seen in 16. They chose to study KNN, SVM with Linear Kernel, Decision Tree (DT), and RF. The dataset was created from three IoT devices from ten minutes of captured network traffic. The models were trained using 85/15 split training and used the Sci-kit learn Python library. The researchers concluded stateless features were more beneficial to classification than stateful features as well as observing RF performing the best and the SVM the worst 15. Similarly in 17, the researchers found the SVM performed the worst in comparison to KNN and RF 18. In the context of binary classification, they were able to achieve accuracies over 99% with a RF. The researchers also found the Radial Basis Function kernel for SVMs worked almost twice as well as the Linear kernel. Following the same trend in 19, researchers utilized Google’s MapReduce as a backbone for network traffic feature extraction, translation, and analysis of changing network features. They tested seven different machine learning based algorithms and RF performed the best with a precision of 0.9994 with SVM trailing far behind at 0.7714. Researchers in 19 set out to test the effectiveness of five different algorithms for intrusion detection. They sourced a dataset from Kaggle and selected 8 different feature vectors to learn from. Training was complete with 80/20 split testing as well as five-fold cross validation referenced in 20. They found RF performs optimally with larger datasets opposed to SVM which regressed in performance. The researchers concluded RFs, Decision Trees, and KNNs could all sufficiently classify and differentiate between normal and attack data and found average accuracy rates observed in the field of IoT to be around 90% for most machine learning algorithms.

3. Methodology

The first step in searching for a dataset was verifying it met the criteria of being published within a year of 2021. After comparing many datasets, our study concluded the ‘Aposemat Iot-23’ dataset sufficiently met all our criteria. Modern datasets are integrable to the quality of results as recently published datasets will include newer devices which current cyber security systems haven’t interacted with before. This leads to untested circumstances and uncovers new exploit surfaces. The study injected the dataset to a Pandas (v1.2.4) framework within the Spyder IDE (v4.2.3) running the Scikit-learn toolkit (v0.24). Training was performed on a PC with the following specifications: Windows 10 (10.0.18363), Intel i7-6800K, 32Gb RAM, and RTX2070 (driver version DCH 466.11). The SVM was tested with both RBF and Linear kernels to account for any variations or performance increases. The study implemented holdout validation with 60/40, 70/30, and 80/20 splits for training the data for both malware detection and intrusion detection to verify the validity of our trained models. The provided Receiver Operating Characteristic (ROC) curves and confusion matrixes are the averaged results from each split.

This article optimized the feature sets for each algorithm for both malware and intrusion detection to get the best performance from each algorithm. To optimize the data, the process started by assessing all the available features in our labeled traffic captures as pictured in Figure 1 and Figure 2. Some features like missed_bytes or resp_pkts were not useful in identifying malware or intrusion-based attacks so they were immediately dismissed for testing. This is due to the fact that these features do not contribute to the probability of identifying a malware or intrusion-based attack with simpler machine learning algorithms. However, note that inclusion of these features may become useful with deep learning algorithms or more advanced datasets that can leverage more data points. Each algorithm shared the same base set of capture features and some algorithms like the RF worked better with an additional feature added. In the case of malware detection, the RF had lower false positive rates and higher accuracies when the features proto and orig_bytes were added (Figure 2). However, the proto feature increased false positive rates significantly in the KNN and SVM as malware and non-malicious programs often use the same UDP and TCP protocols. Additional testing should train with Peer2Peer based traffic and anonymized MAC addresses which has become common practice within the field of cybersecurity.

3.1. Malware Detection

To properly identify malware on the local IoT network, our testing utilized a base feature set of uid, resp_ip_bytes, resp_pkts, orig_ip_pkts, history, resp_bytes, and duration as noted in Figure 1. The uid feature allowed the researchers to track a device across the local network regardless of if the device’s IP changes or techniques to spoof or mask the device’s identity occur. The resp_ip_bytes and resp_bytes sizes are typically very small or zero when malware is present on the network. Contrary to resp_ip_bytes, certain malware increases the orig_ip_pkts total, making this a good identifying metric. As for the history feature, it had minimal to slight effects on lowering false positive rates. Since the results were not negatively impacted, this article kept it as a feature as future testing may utilize it. Finally, the duration is a great feature to use as malware generally has very long connection times. A benign connection may have a connection length of 0.001482 seconds while malware is at 3.151458 seconds, giving a clear flag of suspicious activity.

3.2. Intrusion Detection

Similarly to malware detection, intrusion detection uses the uid to better track the device across the local IoT network. They also share the duration feature to identify suspicious lengths of time indicative of malicious intent. The timestamps (TS) feature is unique to intrusion-based attacks as a vertical or horizontal port scan will have very similar TS from the same device across many IPs or ports on the local network, depending on attack type. To better utilize the TS feature, it needs to be paired with an identifier for cross referencing. In this case, the id.resp_p and if.orig_h features can be used to determine the IP address of the origin device and IP of the destination device. This contributed to higher accuracies as similar timestamps can be compared against host and receiving IP addresses with connection duration to produce significant results.

4. Results and Analysis

This study found the RF algorithm to be the best performing as seen in Table 1. It boasted high accuracies of 92.96% and 92.27% in intrusion and malware detection, respectively, and lowest overall false positive rates. The SVM was found to perform more optimally for malware detection with the Linear kernel and for intrusion detection with the RBF kernel. The KNN was the second-best performing algorithm, following closely behind RF in intrusion detection and malware detection with F1-scores within 3% of the RF and accuracies of 91.48% and 89.80%. With the exclusive features selected in this article, we were able to identify horizontal and vertical port scans as well as malware on an IoT network. The authors of the dataset labeled the traffic as benign or malicious, so this article was able to validate our results for accuracies and false negatives rather than relying on timely lookups that are prone to human error. Our research also concluded that the RF had a more consistent ROC curve across both scenarios compared to the KNN and SVM. The underlying factors to contribute to this are not directly known as the SVM showed irregular inconsistencies, potentially from the different kernels used between malware and intrusion detection as highlighted in 20. Through various testing and validation methods, this study was able to conclude that aging machine learning algorithms are still effective at identifying malware and intrusion detection for IoT devices and networks. Table 1 through 2 exhibit the averaged results for these algorithms.

Table 3 and Table 4 highlight a low-level comparison of the three algorithms tested, all of which have the combined averages of the 60/40, 70/30, and 80/20 data splits.

5. Discussion

As 3, 21 previously found, the RF was the best performing algorithm out of the three tested. The confusion matrices and ROC curves of our results are seen in Figure 3 and Figure 4. The visualization of our results assists in better understandings of the data and its granularity. The ROC curve measures the ratio between the true and false positive rates, with higher ROC values indicating the algorithm can effectively distinguish between positive instances in the data; in the context of our experiment, if a device is benign or malicious. A comparison between the area below the ROC curve for the SVM, KNN and RF can also be seen in Figure 4. As stated before, the numbers present are averages of 60/40, 70/30, and 80/20 data splits to reduce bias.

These results in Table 2, Table 3, and Table 4 support the observations from 18, 22 that the differences in kernel had a large impact on the results and that SVMs perform the worst in intrusion detection. We observed similar results in our experiment as 15, where the authors successfully created a system that can identify malicious or anomalous data in real-time and demonstrated how RFs outperformed the other for algorithms tested. This is also seen in 19 and 23. Similarly to our results, 15 also demonstrated the RFs optimal performance when fed stateless feature sets only. 19 and 23 were also able to accurately identify attack types based upon the data and identify horizontal and vertical port scans with the feature sets, which we successfully replicated. This is observed in the high true positive rate and low false negative scores across all of our algorithms in our confusion matrices. The determination can be safely made that each algorithm is suitable for modern-day malware and intrusion detection, an improved and refined conclusion to results in 24, 25.

6. Conclusion

This article has discussed, as noted in Table 5, and tested the effectiveness of three machine learning algorithms used in IoT cybersecurity for malware and intrusion detection trained on a modern dataset. This study highlighted the importance of maintaining and evolving the cybersecurity protocols for IoT devices and networks to prevent malicious attackers. Some limitations in our study include testing more SVM kernels, such as polynomial kernels, to better optimize SVM results. We believe that future kernel testing may lead to much greater accuracies with the SVM for malware detection. This could also address the aforementioned ROC curve inconsistencies. A small improvement for future studies that may be overlooked is combing datasets to allow for more diversity in networking setups, devices, and protocols to allow for more variables to be tested against. Larger datasets may also more accurately represent the entropy present in real-world implementations. Another limitation is the age of the data within the dataset this article used. Even though the dataset was published in 2020, some of the captures within the dataset date back as far as 2018. This presents the issue of trained models lagging up to three years behind current IoT networks. A three-year window allows for many changes to IoT devices like firmware changes, OS version updates, hardware revisions, and new protocols; all of which can lead to potential exploits requiring constant supervision of cybersecurity measures in place to maintain their effectiveness. Our study explored RFs efficacy with modern IoT data and confirmed that it continues to be the most optimal algorithm to use for IoT cybersecurity. While the KNN and SVM did not perform better than the RF, they can still be effectively used for malware and intrusion detection in today’s IoT networks.

References

[1]  Jayakumar, H., Lee, K., Lee, W. S., Raha, A., Kim, Y., & Raghunathan, V. (2014). Powering the internet of things. In Proceedings of the 2014 international symposium on Low power electronics and design (pp. 375-380).
In article      View Article
 
[2]  R. Dave, “Iot security and authentication schemes based on machine learning,” arXiv preprint arXiv:2109.02695, 2021.
In article      
 
[3]  L. Xiao, X. Wan, X. Lu, Y. Zhang and D. Wu, (2018) “IoT Security Techniques Based on Machine Learning: How Do IoT Devices Use AI to Enhance Security?,” in IEEE Signal Processing Magazine, vol. 35, no. 5, pp. 41-49.
In article      View Article
 
[4]  Agustin Parmisano, Sebastian Garcia, Maria Jose Erquiaga. (2020). Stratosphere Laboratory. A labeled dataset with malicious and benign IoT network traffic. January 22th. https://www.stratosphereips.org/datasets-iot24.
In article      
 
[5]  Hindy, H., Bayne, E., Bures, M., Atkinson, R., Tachtatzis, C., & Bellekens, X. (2020, September). Machine learning based IoT Intrusion Detection System: an MQTT case study (MQTT-IoT-IDS2020 Dataset). In International Networking Conference (pp. 73-84). Springer, Cham.
In article      View Article
 
[6]  Davis, G. (2018). 2020: Life with 50 billion connected devices. 2018 IEEE International Conference on Consumer Electronics (ICCE).
In article      View Article
 
[7]  Gunn, Dylan J. et al. (2019) “Touch-Based Active Cloud Authentication Using Traditional Machine Learning and LSTM on a Distributed Tensorflow Framework.” Int. J. Comput. Intell. Appl. 18 1950022:1-1950022: 16.
In article      View Article
 
[8]  Fang, He, Angie Qi, and Xianbin Wang. (2020). “Fast Authentication and Progressive Authorization in Large-Scale IoT: How to Leverage AI for Security Enhancement.” IEEE Network 34.3 24-29.
In article      View Article
 
[9]  Raza, S., Shafagh, H., Hewage, K., Hummen, R., & Voigt, T. (2013). Lithe: Lightweight Secure CoAP for the Internet of Things. IEEE Sensors Journal, 13(10), 3711-3720.
In article      View Article
 
[10]  Miorandi D, Sicari S, De Pellegrini F, Chlamtac I. (2012). Internet of things: vision, applications and research challenges. Ad Hoc Netw 10(7): 1497.
In article      View Article
 
[11]  T. Kelley and E. Furey, (2018), “Getting Prepared for the Next Botnet Attack: Detecting Algorithmically Generated Domains in Botnet Command and Control,” 2018 29th Irish Signals and Systems Conference (ISSC), Belfast, pp. 1-6.
In article      View Article
 
[12]  Shelton, J., Rice, C., Singh, J., Jenkins, J., Dave, R., Roy, K., & Chakraborty, S. (2018, August). Palm Print Authentication on a Cloud Platform. In 2018 International Conference on Advances in Big Data, Computing and Data Communication Systems (icABCD) (pp. 1-6).
In article      View Article  PubMed
 
[13]  Mason, J., Dave, R., Chatterjee, P., Graham-Allen, I., Esterline, A., & Roy, K. (2020, December). An Investigation of Biometric Authentication in the Healthcare Environment. Array, 8, 100042.
In article      View Article
 
[14]  Kolias, C., Kambourakis, G., Stavrou, A., & Voas, J. (2017). DDoS in the IoT: Mirai and Other Botnets. Computer, 50(7), 80-84.
In article      View Article
 
[15]  Doshi, R., Apthorpe, N., & Feamster, N. (2018). Machine Learning DDoS Detection for Consumer Internet of Things Devices. IEEE Security and Privacy Workshops (SPW).
In article      View Article
 
[16]  Cenedese, A., Zanella, A., Vangelista, L., & Zorzi, M. (2014). Padova Smart City: An urban Internet of Things experimentation. Proceeding of IEEE International Symposium on a World of Wireless, Mobile and Multimedia Networks 2014.
In article      View Article
 
[17]  Alrashdi, Ibrahim, et al. (2019). “Ad-iot: Anomaly detection of iot cyberattacks in smart city using machine learning.” IEEE 9th Annual Computing and Communication Workshop and Conference (CCWC).
In article      View Article
 
[18]  Chhabra, Gurpal Singh, Varinder Pal Singh, and Maninder Singh. (2020). “Cyber forensics framework for big data analytics in IoT environment using machine learning.” Multimedia Tools and Applications 79.23 15881-15900.
In article      View Article
 
[19]  Hasan, M., Milon Islam, M., Islam, I., & Hashem, M. M. A. (2019). Attack and Anomaly Detection in IoT Sensors in IoT Sites Using Machine Learning Approaches. Internet of Things, 100059.
In article      View Article
 
[20]  Moh, M., & Raju, R. (2018) Machine Learning Techniques for Security of Internet of Things (IoT) and Fog Computing Systems. 2018 International Conference on High Performance Computing & Simulation (HPCS).
In article      View Article
 
[21]  Meidan, Y., Bohadana, M., Shabtai, A., Ochoa, M., Tippenhauer, N. O., Guarnizo, J. D., & Elovici, Y. (2017). Detection of unauthorized IoT devices using machine learning techniques. arXiv preprint arXiv:1709.04647.
In article      
 
[22]  Thakkar, A., & Lohiya, R. (2020). A Review on Machine Learning and Deep Learning Perspectives of IDS for IoT: Recent Updates, Security Issues, and Challenges. Archives of Computational Methods in Engineering.
In article      View Article
 
[23]  Rasheed Ahmad, Izzat Alsmadi, Machine learning approaches to IoT security: A systematic literature review , Internet of Things, Volume 14, 2021, 100365, ISSN 2542-6605.
In article      View Article
 
[24]  Strecker S., Van Haaften W., Dave R. (2021). An Analysis of IoT Cyber Security Driven by Machine Learning. In: Kumar S., Purohit S.D., Hiranwal S., Prasad M. (eds) Proceedings of International Conference on Communication and Computational Technologies. Algorithms for Intelligent Systems. Springer, Singapore.
In article      View Article
 
[25]  Siddiqui N., Pryor L., Dave R. (2021). User Authentication Schemes Using Machine Learning Methods—A Review. In: Kumar S., Purohit S.D., Hiranwal S., Prasad M. (eds) Proceedings of International Conference on Communication and Computational Technologies. Algorithms for Intelligent Systems. Springer, Singapore.
In article      View Article
 

Published with license by Science and Education Publishing, Copyright © 2021 Sam Strecker, Rushit Dave, Nyle Siddiqui and Naeem Seliya

Creative CommonsThis work is licensed under a Creative Commons Attribution 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Cite this article:

Normal Style
Sam Strecker, Rushit Dave, Nyle Siddiqui, Naeem Seliya. A Modern Analysis of Aging Machine Learning Based IoT Cybersecurity Methods. Journal of Computer Sciences and Applications. Vol. 9, No. 1, 2021, pp 16-22. http://pubs.sciepub.com/jcsa/9/1/2
MLA Style
Strecker, Sam, et al. "A Modern Analysis of Aging Machine Learning Based IoT Cybersecurity Methods." Journal of Computer Sciences and Applications 9.1 (2021): 16-22.
APA Style
Strecker, S. , Dave, R. , Siddiqui, N. , & Seliya, N. (2021). A Modern Analysis of Aging Machine Learning Based IoT Cybersecurity Methods. Journal of Computer Sciences and Applications, 9(1), 16-22.
Chicago Style
Strecker, Sam, Rushit Dave, Nyle Siddiqui, and Naeem Seliya. "A Modern Analysis of Aging Machine Learning Based IoT Cybersecurity Methods." Journal of Computer Sciences and Applications 9, no. 1 (2021): 16-22.
Share
[1]  Jayakumar, H., Lee, K., Lee, W. S., Raha, A., Kim, Y., & Raghunathan, V. (2014). Powering the internet of things. In Proceedings of the 2014 international symposium on Low power electronics and design (pp. 375-380).
In article      View Article
 
[2]  R. Dave, “Iot security and authentication schemes based on machine learning,” arXiv preprint arXiv:2109.02695, 2021.
In article      
 
[3]  L. Xiao, X. Wan, X. Lu, Y. Zhang and D. Wu, (2018) “IoT Security Techniques Based on Machine Learning: How Do IoT Devices Use AI to Enhance Security?,” in IEEE Signal Processing Magazine, vol. 35, no. 5, pp. 41-49.
In article      View Article
 
[4]  Agustin Parmisano, Sebastian Garcia, Maria Jose Erquiaga. (2020). Stratosphere Laboratory. A labeled dataset with malicious and benign IoT network traffic. January 22th. https://www.stratosphereips.org/datasets-iot24.
In article      
 
[5]  Hindy, H., Bayne, E., Bures, M., Atkinson, R., Tachtatzis, C., & Bellekens, X. (2020, September). Machine learning based IoT Intrusion Detection System: an MQTT case study (MQTT-IoT-IDS2020 Dataset). In International Networking Conference (pp. 73-84). Springer, Cham.
In article      View Article
 
[6]  Davis, G. (2018). 2020: Life with 50 billion connected devices. 2018 IEEE International Conference on Consumer Electronics (ICCE).
In article      View Article
 
[7]  Gunn, Dylan J. et al. (2019) “Touch-Based Active Cloud Authentication Using Traditional Machine Learning and LSTM on a Distributed Tensorflow Framework.” Int. J. Comput. Intell. Appl. 18 1950022:1-1950022: 16.
In article      View Article
 
[8]  Fang, He, Angie Qi, and Xianbin Wang. (2020). “Fast Authentication and Progressive Authorization in Large-Scale IoT: How to Leverage AI for Security Enhancement.” IEEE Network 34.3 24-29.
In article      View Article
 
[9]  Raza, S., Shafagh, H., Hewage, K., Hummen, R., & Voigt, T. (2013). Lithe: Lightweight Secure CoAP for the Internet of Things. IEEE Sensors Journal, 13(10), 3711-3720.
In article      View Article
 
[10]  Miorandi D, Sicari S, De Pellegrini F, Chlamtac I. (2012). Internet of things: vision, applications and research challenges. Ad Hoc Netw 10(7): 1497.
In article      View Article
 
[11]  T. Kelley and E. Furey, (2018), “Getting Prepared for the Next Botnet Attack: Detecting Algorithmically Generated Domains in Botnet Command and Control,” 2018 29th Irish Signals and Systems Conference (ISSC), Belfast, pp. 1-6.
In article      View Article
 
[12]  Shelton, J., Rice, C., Singh, J., Jenkins, J., Dave, R., Roy, K., & Chakraborty, S. (2018, August). Palm Print Authentication on a Cloud Platform. In 2018 International Conference on Advances in Big Data, Computing and Data Communication Systems (icABCD) (pp. 1-6).
In article      View Article  PubMed
 
[13]  Mason, J., Dave, R., Chatterjee, P., Graham-Allen, I., Esterline, A., & Roy, K. (2020, December). An Investigation of Biometric Authentication in the Healthcare Environment. Array, 8, 100042.
In article      View Article
 
[14]  Kolias, C., Kambourakis, G., Stavrou, A., & Voas, J. (2017). DDoS in the IoT: Mirai and Other Botnets. Computer, 50(7), 80-84.
In article      View Article
 
[15]  Doshi, R., Apthorpe, N., & Feamster, N. (2018). Machine Learning DDoS Detection for Consumer Internet of Things Devices. IEEE Security and Privacy Workshops (SPW).
In article      View Article
 
[16]  Cenedese, A., Zanella, A., Vangelista, L., & Zorzi, M. (2014). Padova Smart City: An urban Internet of Things experimentation. Proceeding of IEEE International Symposium on a World of Wireless, Mobile and Multimedia Networks 2014.
In article      View Article
 
[17]  Alrashdi, Ibrahim, et al. (2019). “Ad-iot: Anomaly detection of iot cyberattacks in smart city using machine learning.” IEEE 9th Annual Computing and Communication Workshop and Conference (CCWC).
In article      View Article
 
[18]  Chhabra, Gurpal Singh, Varinder Pal Singh, and Maninder Singh. (2020). “Cyber forensics framework for big data analytics in IoT environment using machine learning.” Multimedia Tools and Applications 79.23 15881-15900.
In article      View Article
 
[19]  Hasan, M., Milon Islam, M., Islam, I., & Hashem, M. M. A. (2019). Attack and Anomaly Detection in IoT Sensors in IoT Sites Using Machine Learning Approaches. Internet of Things, 100059.
In article      View Article
 
[20]  Moh, M., & Raju, R. (2018) Machine Learning Techniques for Security of Internet of Things (IoT) and Fog Computing Systems. 2018 International Conference on High Performance Computing & Simulation (HPCS).
In article      View Article
 
[21]  Meidan, Y., Bohadana, M., Shabtai, A., Ochoa, M., Tippenhauer, N. O., Guarnizo, J. D., & Elovici, Y. (2017). Detection of unauthorized IoT devices using machine learning techniques. arXiv preprint arXiv:1709.04647.
In article      
 
[22]  Thakkar, A., & Lohiya, R. (2020). A Review on Machine Learning and Deep Learning Perspectives of IDS for IoT: Recent Updates, Security Issues, and Challenges. Archives of Computational Methods in Engineering.
In article      View Article
 
[23]  Rasheed Ahmad, Izzat Alsmadi, Machine learning approaches to IoT security: A systematic literature review , Internet of Things, Volume 14, 2021, 100365, ISSN 2542-6605.
In article      View Article
 
[24]  Strecker S., Van Haaften W., Dave R. (2021). An Analysis of IoT Cyber Security Driven by Machine Learning. In: Kumar S., Purohit S.D., Hiranwal S., Prasad M. (eds) Proceedings of International Conference on Communication and Computational Technologies. Algorithms for Intelligent Systems. Springer, Singapore.
In article      View Article
 
[25]  Siddiqui N., Pryor L., Dave R. (2021). User Authentication Schemes Using Machine Learning Methods—A Review. In: Kumar S., Purohit S.D., Hiranwal S., Prasad M. (eds) Proceedings of International Conference on Communication and Computational Technologies. Algorithms for Intelligent Systems. Springer, Singapore.
In article      View Article