- Title
- Modelling internet network intrusion detection in smart city ecosystems
- Creator
- Mfenguza, Wandisa
- Subject
- Ecosystem management
- Subject
- Smart cities
- Date Issued
- 2021-05
- Date
- 2021-05
- Type
- Master's theses
- Type
- text
- Identifier
- http://hdl.handle.net/10353/22501
- Identifier
- vital:52382
- Description
- Smart city systems are intended to enhance the lives of citizens through the design of systems that promote resource efficiency and the real-time provisioning of resources in cities. The benefits offered by smart cities include the use of internet of things (IoT) sensors to gather useful data such as power demand to inhibit blackouts and the average speed of vehicles to alleviate traffic congestion. Nonetheless, earlier studies have indicated a substantial increase in cyber-security issues due to the increase in the deployment of smart city ecosystems. Consequently, IoT cyber-security is recognised as an area that requires crucial scrutiny. This study begins by investigating the current state of intrusion detection in smart city ecosystems. Current intrusion detection frameworks lack the capability to operate under extremely limiting settings such as conditions of low processing power and fast response times. Moreover, the study also identifies that, despite intrusion detection being a highly researched thematic area, a plethora of previous studies tend to propose intrusion detection frameworks that are more suitable for traditional computer networks rather than wireless sensor networks (WSNs) which consist of heterogeneous settings with diverse devices and communication protocols. Subsequently, this study developed two candidate deep learning models, namely a convolutional neural network (CNN) and a long short-term memory (LSTM) network and presents evidence on their robustness and predictive power. Results have indicated that, unlike the CNN model, the LSTM model can quickly converge and offer high predictive power without the vigorous application of regularisation techniques. The proposed LSTM classification model obtained a remarkable 100% in detection rates and further reported 0% in false alarm and false negative rates. This study gives a broad overview of the current state of intrusion detection mechanisms for smart city ecosystems to guide future studies. The study also demonstrates that existing intrusion detection systems (IDSs) can be enhanced through the development of more robust and lightweight models that offer high detection rates and minimal false alarm rates to prevent security risks in smart city ecosystems to ensure sustainable and safe smart cities.
- Description
- Thesis (MSc) -- Faculty of Science and Agriculture, 2021
- Format
- computer
- Format
- online resource
- Format
- application/pdf
- Format
- 1 online resource (111 pages)
- Format
- Publisher
- University of Fort Hare
- Publisher
- Faculty of Science and Agriculture
- Language
- English
- Rights
- rights holder
- Rights
- All Rights Reserved
- Rights
- Open Access
- Hits: 420
- Visitors: 450
- Downloads: 66
Collections
UFH Department of Computer Science
| Thumbnail | File | Description | Size | Format | |||
|---|---|---|---|---|---|---|---|
| View Details Download | SOURCE1 | Masters Disssertation - 201308763 Mfenguza W (MSc Computer Science).pdf | 1 MB | Adobe Acrobat PDF | View Details Download |