Fluid Antennas for Non-Monotonic THz-WPT With Channel Estimation Errors

In this study, we explore a terahertz (THz) wireless power transfer (WPT) system. Since traditional Schottky diodes may be inefficient for THz rectification, we focus on a resonant tunneling diode (RTD)-based rectifier. Such rectifiers are generally characterized by a non- monotonic energy harvesting (EH) behavior, where an increase in input power may not necessarily correspond to higher levels of harvested power. To this end, we propose the exploitation of fluid antennas (FAs) to align the non- monotonic rectification of RTDs with favorable channel realizations. More specifically, a generalized piecewise linear function is adopted in order to approximate the instantaneous input-output power relationship. Based on this, we derive an analytical framework in terms of the energy outage probability and the average harvested power for three FA port selection policies, namely (i) the input-based selection, (ii) the harvesting-based selection, and (iii) the random selection, each corresponding to different performance. The impact of channel estimation errors on the port selection strategies is then examined, revealing a trade-off between the number of estimated ports and harvested power. To further improve EH performance, we introduce a new receiver architecture featuring a power splitter, multiple RTD-based EH circuits and a DC combiner. Based on this, we provide two dynamic power splitting schemes tailored to avoid inefficient rectification regions. Numerical results which validate our analysis, reveal a novel utilization of FAs, stemming from the alignment of the port selection process with the non- monotonic harvesting characteristics. Finally, our findings demonstrate that the proposed receiver architecture can yield significant gains in harvested power.