Selected Topic: Regulation for Trust in Digital Environments
Cloud users and providers. Users are assured that their privacy requirements are met by the provider, while providers have means to implement such privacy requirements.
Internet of Things adoption is growing steadily and its integration with Cloud Computing is increasing the number of devices connected to the Internet. IoT devices, that are responsible for communication and actuation on user information, that are often sensitive, are exposed to several types of attacks, which may expose all other interconnected devices. Several proposals approach the security and privacy in the Internet of Things, most of then centralize the security protocols in a gateway device, which is responsible for aggregating the protection IoT data and sending it to the cloud. This approach brings several benefits, one very important is that it is possible to enforce security and privacy regardless of IoT networks technologies, which means that the protocol can be used in any situation.
In our work, we propose an architecture for privacy and the integration of the Internet of Things and Cloud Computing, adapting and moving the privacy and security mechanisms from the gateway to the IoT devices. This approach is aligned with efforts from several standardization organizations, that, in the last years, propose several communication standards for the Internet of Things with the goal of decreasing the fragmentation among current solutions.
Based in User-driven Privacy Enforcement for Cloud-based Services in the Internet of Things (UPECSI), the proposed architecture stores IoT data encrypted at the cloud platform, in this way only authorized entities (user and cloud services) have access to it, improving assurance of user privacy requirements. The approach of implementing security schemes at the IoT devices improves the architecture’s fault-tolerance, since it removes a single point of failure. It also improves the overall security of the system, since there is no need of a component responsible for all user messages, which could impair the security properties of the system once compromised by a successful attack.
We called the proposed architecture as: PROTeCt: Privacy aRquitecture for integratiOn of internet of Things and Cloud computing. In our view, user’s IoT networks will send data to the cloud to provide useful services to the user. When uploading data, IoT devices encrypt it with a symmetric key, which only authorized services have access. Keys are renewed periodically, enabling the addition and removal of services access to the data. Services are implemented through a Privacy Development Language (PDL), which requires to inform the actions taken with user data and enables users to enable and disable specific services features according to user’s privacy requirements. A Trusted Third Party is responsible for auditing cloud services. We also propose Enhanced PROTeCt (E-PROTeCt), that decreases processing and transmission overhead by applying security schemes at the application layer instead of transport layer. This approach avoids encrypting data twice, since it must be encrypted for storage.
PROTeCt and E-PROTeCt were evaluated through analytical analysis and simulation, results were compared with UPECSI and a network without security. We derived equations to determine the overhead in processing: how much data was encrypted and how much data was sent. We conducted simulations using the ns-3 network simulator, we used experimentally acquired times for the cryptographic operations at the IoT devices and at the gateway. To show that our approach was suitable for even severely constrained devices, we simulated the use of a Micaz Mote, a device with an 8-bit ATmega128L microcontroller, 128KB RAM and 512KB ROM that communicates at 250 kbps. Several simulation scenarios where executed, varying the number of devices from 4 to 160 and the interval of generated data from 15 seconds to 5 minutes. Delay and energy consumption were evaluated, results show that delay overhead is around 3,5 ms for PROTeCt, and 1,75 ms for E-PROTeCt. Energy consumption was evaluated through lifetime comparison, considering a battery of 2600 mAh. In this case results show an overhead around 21,80% for PROTeCt and 15,51% for E-PROTeCt.
Results show that it is possible to implement security and privacy schemes at IoT devices instead of relying on a single point of failure to enforce those properties for the entire network. It is also important to notice that the presented overhead is regarding an IoT network without any security, which turns security from gateway to cloud irrelevant. To provide means of enforcing user privacy requirements in a cloud environment is paramount for its wide adoption. The Internet of Things amplifys this requirement, since it significantly increases the amount of data sent to the cloud, showing the importance of this work.