Evaluating Machine Learning Methods for Intrusion Detection in IoT

Cyber-attacks in IoT enabled devices have grown at an alarming rate since the start of the Covid-19 pandemic due to cyber physical digital transformation enabled through widespread deployment of low cost sensor embedded IoT devices in consumer and industrial IOT, as well as increase in computing power. Consequently, this adoption trend had led to 1.51 billion breaches on IoT devices during the first half of 2021 alone. This highlights the critical importance of being prepared for IoT vulnerabilities (IoT manufacturing and deployment sector) and attacks (malicious actors). In this respect machine learning (ML) especially deep learning (DL) strategies has emerged as the preferred methods to secure IoT devices from attacks. In this paper, we propose three deep learning algorithms for IoT intrusion detection based on mapping of IoT attacks to ML/DL methods. Our paper thus provides two contributions. First, we present a model that maps extant research on the application of ML/DL to specific IoT attacks. Second, through an optimal selection of the mapping, we present three algorithms (naïve Bayes, convolutional neural network and autoencoder) for detection of intrusion in IoT attacks. This provides a review of research opportunities and research gaps in the IoT IDS domain. © 2022 ACM.

Application of Consumptive Metrics to Measure Internship Alignment

Internships aim at transitioning students from the academic environment (academic learning at the university) to a professional work environment (industry practice). Our paper aims to objectively evaluate the alignment of learning with practice based on the internship program conducted in Term 1, 2020 (pre-Covid), for our undergraduate students at the College of Technology Innovation studying in the bachelor's program for Computer Science and Information Systems. In order to measure the alignment, from a theoretical perspective, we adopted the framework of Kirkpatrick, which provides a set of "consumptive metrics"for evaluating the learning resources consumed in education and training, using the constructs 'reaction' (how the learners feel, including their personal reactions to the internship training) and 'learning' (measuring the knowledge, skills, or attitudes acquired as a direct result of the training, including mapping to their courses). Using 36 internship student reports collected over a single semester (in which students spent 8 weeks onsite at various organizations in the United Arab Emirates) as a sample for this study, we measured internship results in terms of the learning resources consumed during the internship experience using consumptive metrics to observe its alignment with practice. The results of the study allow academics to reinforce strong areas and improve areas of concern to better align learning with practice. © 2021 IEEE.

Replacing Human Input in Spam Email Detection Using Deep Learning

The Covid-19 pandemic has been a driving force for a substantial increase in online activity and transactions across the globe. As a consequence, cyber-attacks, particularly those leveraging email as the preferred attack vector, have also increased exponentially since Q1 2020. Despite this, email remains a popular communication tool. Previously, in an effort to reduce the amount of spam entering a users inbox, many email providers started to incorporate spam filters into their products. However, many commercial spam filters rely on a human to train the filter, leaving a margin of risk if sufficient training has not occurred. In addition, knowing this, hackers employ more targeted and nuanced obfuscation methods to bypass in-built spam filters. In response to this continued problem, there is a growing body of research on the use of machine learning techniques for spam filtering. In many cases, detection results have shown great promise, but often still rely on human input to classify training datasets. In this study, we explore specifically the use of deep learning as a method of reducing human input required for spam detection. First, we evaluate the efficacy of popular spam detection methods/tools/techniques (freeware). Next, we narrow down machine learning techniques to select the appropriate method for our dataset. This was then compared with the accuracy of freeware spam detection tools to present our results. Our results showed that our deep learning model, based on simple word embedding and global max pooling (SWEM-max) had higher accuracy (98.41%) than both Thunderbird (95%) and Mailwasher (92%) which are based on Bayesian spam filtering. Finally, we postulate whether this improvement is enough to accept the removal of human input in spam email detection. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

IoTVT Model: A Model Mapping IoT Sensors to IoT Vulnerabilities and Threats

The Internet of Things (IoT), which has accelerated the digital transformation technology revolution, has enabled cyber-physical digital transformation strategies and accelerated business automation. In a Covid-19 related Harvard Business Review study, 95 per cent of executives agreed that digital transformation strategies had become increasingly important. This highlights the critical importance of being prepared for IoT vulnerabilities and attacks. Mapping IoT devices to identify their vulnerabilities and attack allows academics and practitioners to identify, analyze, and mitigate IoT-related concerns. In this paper, we categorize IoT sensors, their IoT related vulnerabilities, and the IoT attacks that affect them and propose a model that maps the relationships among them. Our model provides valuable insights into IoT attack vectors and associated vulnerabilities with consumer IoT devices. © 2021 IEEE.