Résumé : This thesis investigates spatial data focusing (SDF) as a means of performing wireless physical-layer geocasting, i.e. location-based multicasting or geographically-confined broadcasting. This novel approach can aid in providing location-based services and messaging to large groups of mobile devices that exist in emerging internet-of-things frameworks for smart cities, industries, healthcare, etc., providing users with information that is related or contextualized to their geographical location. It addresses and avoids privacy concerns that exist in conventional location-based services, where users are required to disclose their location. In addition, it overcomes node self-localization requirements and the challenging balance between overhead, scalability, and delivery rate that exist in network-layer geocast routing algorithms. Most importantly, it succeeds in increasing precision, reducing array size, and minimizing complexity -- the most crucial conditions in making physical-layer geocasting an attractive scheme -- compared to conventional beamforming-based power focusing approaches. Within the SDF framework, it additionally addresses two fundamental shortcomings. That is, (i) a limitation to focusing in the angular domain only or, equivalently, the inability for range-domain focusing and (ii) a severe sensitivity to multipath propagation that jeopardizes correct operation outside hypothetical free space channels. They are overcome by designing two novel SDF architectures that exploit multi-frequency transmission resources in an orthogonal frequency-division multiplexing (OFDM) and frequency diverse array (FDA) framework. Additionally, an experimental proof-of-concept SDF architecture is developed that demonstrates its practical achievability as a novel geocasting technique.