Lightweight cryptography and authentication protocols for secure communications between resource-limited devices and wireless sensor networks: evaluation and implementation

摘要

This dissertation examines the theoretical context for the security of wireless communication between ubiquitous computing devices and presents an implementation that addresses this need. The number of Resource-Limited Wireless Devices utilized in many areas of the IT industry is growing rapidly. Some of the applications of these devices pose real security threats that can be addressed using authentication and cryptography.udMany of the available authentication and encryption software solutions are predicated on the availability of ample processing power and memory. These demands cannot be met by the majority of ubiquitous computing devices, thus there is a need to apply lightweight cryptography primitives and lightweight authentication protocols that meet these demands in any application of security to devices with limited resources.udThe analysis of the lightweight solutions is divided into two major sections: Lightweight Authentication Protocols and Lightweight Encryption Algorithms. Further sections of this work describe the proposed prototype‟s Wireless Sensor Network including a study of its limitations.udA number of protocols in the field of Authentication and in the field of Encryption are analyzed. The Gossamer Authentication Protocol and the Scalable Encryption Algorithm (SEA) are chosen as the basis of prototype implementation in the C-language on a development platform of the 8051-compatible Nordic Semiconductor nRF9E5 microcontroller. A security framework is developed that combines the attributes of the Gossamer protocol and the SEA to provide an implementation of inter-device security. The Gossamer Protocol is additionally used as a means of exchanging session keys for use with the SEA encryption protocol. A brief performance analysis of the prototype running on the nRF9E5 microcontroller is provided by way of conclusion. The resuls of the software implementation of the Gossamer were unsatisfactory both in terms of the code space needs (approximately 1700 bytes excluding shared libraries) and the execution time (almost 150 milliseconds). In contrast, the SEA implementation‟s results were satisfactory above expectations with the code space requirements smaller than 600 bytes (excluding shared libraries) and the performance of 27 milliseconds per one 96-bit block of data.