Abstract
The Internet of Things ecosystem enables massive sensor data acquisition, transmission, and storage for diverse smart monitoring, process automation, and control services. It has revolutionized many areas such as body area networks, industrial robots, smart homes, smart vehicles, and so forth. Despite having many advantages, the ever-increasing data generation and huge number of devices bring many critical challenges such as increasing pressure on communication and storage infrastructure, end-to-end information secrecy, resource-constrained sensors, and data analytics, in the IoT ecosystem.
This PhD project aims to design novel lightweight schemes to provide data compression and encryption for resource-constrained IoT devices, which could also provide energy-saving, communication cost reduction, and end-to-end information secrecy.
The first part of the dissertation covers the security analysis of Compressive Sensing (CS) based joint compression and information secrecy schemes. It is shown that the CS-based computationally secure scheme is vulnerable to ciphertext-only attack. A novel transformation is proposed to achieve perfect secrecy using CS by concealing signal energy. Furthermore, a novel energy concealment scheme is proposed. Prototypes of the variants of the energy concealment EC scheme are implemented in a resource-constrained device. Each variant's performance and security analysis is compared with the state-of-the-art scheme. The EC scheme can achieve energy savings in transmission by 52% to 68% compared with the advanced encryption standard solution. Moreover, important questions of reusability and indistinguishability of binary sensing matrices are addressed in the dissertation.
In the second part of the dissertation, a unified framework to combine compression, encryption, and error recovery in the CS encoding process is proposed using projection-based encoding. It is proved that compression, encryption, and error recovery can be combined in a unified framework. Based on this unified framework, two novel schemes, EfficieNt seCure eRror-robUST (ENCRUST) and Lightweight ENCRUST (L-ENCRUST), are proposed. Moreover, for each scheme, the security analysis is performed, and the prototype is implemented in a TelosB mote. ENCRUST, and L-ENCRUST can reduce total energy consumption by 12% and 26%, respectively, compared with the state-of-the-art solution.
Finally, multi-class encryption scenarios are studied, and a novel Multi-class Privacy-preserving Cloud Computing (MPCC) scheme is proposed which provides encryption of plaintext for two different users using a single encoding process. Moreover, statistical decryption and data anonymization variants of MPCC are proposed for electrocardiogram signals and images by considering the signal characteristics.
This PhD project aims to design novel lightweight schemes to provide data compression and encryption for resource-constrained IoT devices, which could also provide energy-saving, communication cost reduction, and end-to-end information secrecy.
The first part of the dissertation covers the security analysis of Compressive Sensing (CS) based joint compression and information secrecy schemes. It is shown that the CS-based computationally secure scheme is vulnerable to ciphertext-only attack. A novel transformation is proposed to achieve perfect secrecy using CS by concealing signal energy. Furthermore, a novel energy concealment scheme is proposed. Prototypes of the variants of the energy concealment EC scheme are implemented in a resource-constrained device. Each variant's performance and security analysis is compared with the state-of-the-art scheme. The EC scheme can achieve energy savings in transmission by 52% to 68% compared with the advanced encryption standard solution. Moreover, important questions of reusability and indistinguishability of binary sensing matrices are addressed in the dissertation.
In the second part of the dissertation, a unified framework to combine compression, encryption, and error recovery in the CS encoding process is proposed using projection-based encoding. It is proved that compression, encryption, and error recovery can be combined in a unified framework. Based on this unified framework, two novel schemes, EfficieNt seCure eRror-robUST (ENCRUST) and Lightweight ENCRUST (L-ENCRUST), are proposed. Moreover, for each scheme, the security analysis is performed, and the prototype is implemented in a TelosB mote. ENCRUST, and L-ENCRUST can reduce total energy consumption by 12% and 26%, respectively, compared with the state-of-the-art solution.
Finally, multi-class encryption scenarios are studied, and a novel Multi-class Privacy-preserving Cloud Computing (MPCC) scheme is proposed which provides encryption of plaintext for two different users using a single encoding process. Moreover, statistical decryption and data anonymization variants of MPCC are proposed for electrocardiogram signals and images by considering the signal characteristics.
| Original language | English |
|---|
| Publisher | Århus Universitet |
|---|---|
| Publication status | Published - Jun 2022 |