Ahmed Mohamed Ali Moussa Eid

• Accurate and efficient simulation of body-antenna interactions is identified as a vital requirement for the design and optimization of wireless systems that operate near the human body. To this end, efficient simulation methods with high accuracy are developed for two important and diverse applications. First, speech sensing is considered, whose aim is to sense and track the process of human speech, requiring information about the position and movement of lips, tongue, glottis, etc., potentially improving synthetic speech production, speech pathology and therapy, and speech recognition. For this purpose, a simplified FDTD model of the head-vocal tract is used which is efficient enough to be simulated on a desktop PC. To improve efficiency of the speech-sensing application, a mode-matching waveguide model is developed. Comparing simulation times, the proposed method is orders of magnitude more efficient than direct simulation with the FDTD solver. Next, the FDTD model is used to design and optimize an antenna that enhances coupling of propagating waves to mouth and throat. To test the accuracy of the modeling procedures and explore the feasibility of speech sensing, a prototype speech sensing system is built and deployed. A speech recognition experiment is also performed as a proof-of-concept to motivate the idea of speech sensing. As a second target application, modeling for body area networking is presented. This application focuses on the shadowing aspect of the body which is most challenging impediment to the communications. An efficient two-dimensional lossy dielectric cylinder model for the body is adopted and extended to arbitrary polarization and arbitrary cross section using a surface-based method of moments solution. Additionally, measurements with short dipoles are performed in a compact anechoic chamber and compared with simulations to illustrates utility and accuracy of the models.