Search

This thesis is an alternative format submission comprising a set of publications and a comprehensive literature review, an introduction, and a conclusion. Continuous compliance with data protection legislation on many levels in the Industrial Internet of Things (IIoT) is a significant challenge. Automated continuous compliance should also consider adaptable security compliance management for multiple users. The IIoT should automate compliance with corporate rules, regulations, and regulatory frameworks for industrial applications. Thus, this thesis aims to improve continuous compliance by introducing an edge-server architecture which incorporates searchable encryption with multi-authority access to provide access to useful data for various stakeholders in the compliance domain. In this thesis, we propose an edge lightweight searchable attribute-based encryption system (ELSA). The ELSA system leverages cloud-edge architecture to improve search time beyond a previous state-ofthe-art encryption solution. The main contributions of the first paper are as follows. First, we npresent an untrusted cloud and trusted edge architecture that processes data efficiently and optimises decision-making in the IIoT context. Second, we enhanced the search performance over the current state-of-the-art (LSABE-MA) regarding order of magnitude. We achieved this enhancement by storing keywords only on the trusted edge server and introducing a query optimiser to achieve better-than-linear search performance. The query optimiser uses k-means clustering to improve the efficiency of range queries, removing the need for a linear search. As a result, we achieved higher performance without sacrificing result accuracy. In the second paper, we extended ELSA to illustrate the correlation between the number of keywords and ELSA performance. This extension supports annotating records with multiple keywords in trapdoor and record storage and enables the record to be returned with single keyword queries. In addition, the experiments demonstrated the scalability and efficiency of ELSA with an increasing number of keywords and complexity. Based on the experimental results and feedback received from the publication and presentation of this work, we published our third technical paper. In this paper, we improved ELSA by minimising the lookup table size and summarising the data records by integrating machine-learning (ML) methods suitable for execution at the edge. This integration removes records of unnecessary data by evaluating added value to further processing. This process results in the minimisation of the lookup table size, the cloud storage, and the network traffic, taking full advantage of the edge architecture benefits. We demonstrated the mini-ELSA expanded method on two well-known IIoT datasets. Our results reveal a reduction of storage requirements by > 21% while improving execution time by > 1.39× and search time by > 50% and maintaining an optimal balance between prediction accuracy and space reduction. In addition, we present the computational complexity analysis that reinforces these experimental results.

The Industrial Internet of Things (IIoT) plays a powerful role in smart manufacturing by performing real-time analysis for large volumes of data. In addition, IIoT systems can monitor several factors, such as data accuracy, network bandwidth and operations latency. To perform these operations securely and in a privacy-preserving manner, one solution is to use cryptographic primitives. However, most cryptographic solutions add performance overhead causing latency. In this paper, we propose an Edge Lightweight Searchable Attribute-based encryption system (ELSA). ELSA leverages the cloud-edge architecture to improve search time beyond the state-of-the-art. The main contributions of this paper are as follows. First, we present an untrusted cloud/trusted edge architecture, which optimises the efficiency of data processing and decision making in the IIoT context. Second, we enhance search performance over current state-of-the-art (LSABE-MA) by an order of magnitude. We achieve this by improving the organisation of the data to provide better than linear search performance. We leverage the edge server to cluster data indices by keyword and introduce a query optimiser. The query optimiser uses k-means clustering to improve the efficiency of range queries, removing the need for linear search. In addition, we achieve this without sacrificing accuracy over the results.