A Lightweight Authentication Architecture for Unsupervised Internet of Things (IoT) in Smart Home Applications

ABSTRACT

The Smart Home environment is made up of different objects that have sensing capabilities

and have the potential to interact with each other seamlessly. This brings a lot of

convenience to the control and monitoring of the surroundings around the home

environment. This reality is brought about as a result of the Internet of Things (IoT)

phenomenon. The potential benefits presented by IoT technologies around the Smart Home

environment can and are hampered by security issues that are yet to be resolved both at

the perception layer and the transmission layer.

The need to secure data collected around the home environment and the exchange of such

data among the smart objects is of paramount importance. The general limitation that

things in the Internet of Things suffer from is that of computational power and storage

space. Resource constrained devices hinder the application of robust security solutions that

conventional networking environment devices enjoy hence the need to look at the suitable

solutions that meet the resource basis of things in an optimal way. To realise this objective,

this research employed a constructivist paradigm, which guided the design of an artefact

that was tested under the guided framework of the design science research approach.

The focus on authentication as a security dimension has been motivated by its interweaved

nature into other security pillars. Authentication proves to be a primary security key

window in that if it fails to detect unauthorised access, all other security loopholes are

opened in the entire networked environment.

A simulated Smart Home environment that modelled critical application requirements for

Assisted Ambient Living (AAL) spaces and Energy Saving Solutions (ESS) was used to evaluate

the proposed lightweight authentication architecture’s efficiency, which was tested against

existing similar solutions around the same functionality. The lightweight authentication

architecture presented in this submission was tested using the SCYTHER tool, which allowed

verification, falsification and security testing by checking on various classes of attacks and

possible architecture behaviour. The architecture turned out secure for tested insider,

impersonation, replay and man-in-middle attacks, which were considered ideal as guided by

the Dolev-Yao model

The contribution of this research is its pragmatic approach to the security design for

constrained things in IoT that can operate with little to no human intervention – hence

unsupervised. Key findings from this work highlight two important aspects for proper

security advancement, which are identity management of things in the IoT space and the

scalability of using agent based models to reduce resource demands at the device level.

As an envisaged current and future relevance of this work, it may inform the security design

of authentication solutions in IoT application environments in ad hoc personal area network

setups and feed into the bigger vision of smart cities.