Enhancing the security of wireless sensor network based home automation systems
thesisposted on 22.02.2010, 13:43 by Khusvinder Gill
Home automation systems (HASs)seek to improve the quality of life for individuals through the automation of household devices. Recently, there has been a trend, in academia and industry, to research and develop low-cost Wireless Sensor Network (WSN) based HASs (Varchola et al. 2007). WSNs are designed to achieve a low-cost wireless networking solution, through the incorporation of limited processing, memory, and power resources. Consequently, providing secure and reliable remote access for resource limited WSNs, such as WSN based HASs, poses a significant challenge (Perrig et al. 2004). This thesis introduces the development of a hybrid communications approach to increase the resistance of WSN based HASs to remote DoS flooding attacks targeted against a third party. The approach is benchmarked against the dominant GHS remote access approach for WSN based HASs (Bergstrom et al. 2001), on a WSN based HAS test-bed, and shown to provide a minimum of a 58.28%, on average 59.85%, and a maximum of 61.45% increase in remote service availability during a DoS attack. Additionally, a virtual home incorporating a cryptographic based DoS detection algorithm, is developed to increase resistance to remote DoS flooding attacks targeted directly at WSN based HASs. The approach is benchmarked against D-WARD (Mirkovic 2003), the most effective DoS defence identified from the research, and shown to provide a minimum 84.70%, an average 91.13% and a maximum 95.6% reduction in packets loss on a WSN based HAS during a DoS flooding attack. Moreover, the approach is extended with the integration of a virtual home, hybrid communication approach, and a distributed denial of defence server to increase resistance to remote DoS attacks targeting the home gateway. The approach is again benchmarked against the D-WARD defence and shown to decrease the connection latency experienced by remote users by a minimum of 90.14%, an average 90.90%, and a maximum 91.88%.
- Computer Science