The low-cost 'THz Torch’ technology, which exploits the thermal infrared spectrum (ca. 10 to 100 THz), was recently introduced to provide secure low data rate communications links across short ranges. In this thesis, the channel model for 'THz Torch’ wireless communications links is redeveloped from a thermodynamics perspective. Novel optimization-based channel estimators are also proposed to calibrate parameters in the channel model. Based on these theoretical advances, a cognitive 'THz Torch’ receiver, which combines conventional digital communications with state-of-the-art deep learning techniques, is presented to achieve cognitive synchronization and demodulation. The newly reported 'THz Torch’ wireless link is capable of bypassing the thermal time constant constraints normally associated with both the thermal emitter and sensor, allowing truly asynchronous data transfer with direct electronic modulation. Experimental results obtained in both laboratory environments and field trials demonstrate step-change improvements in channel range, bit rate, bit error rate and demodulation speed.

This work represents a paradigm shift in modulation-demodulation with a thermal-based physical layer and offers a practical solution for implementing future ubiquitous secure 'THz Torch’ wireless communications links. The cognitive receiver concept also has wide-ranging implications for future communications and sensor technologies, making them more resilient when operating in harsh environments.