Recursive Waterfilling for Wireless Links With Energy Harvesting Transmitters

Wireless Energy Harvesting Using Signals From Multiple Fading Channels

“…In this case, one has the input-output relationship of the energy harvester as [4] - [9] f (x) = ηx (7) where η is the conversion efficiency of the energy harvester and x is the input power. An important assumption here is that the conversion efficiency is a constant that is independent of the input power.…”

“…measures for best performance [4] - [6], there is still great uncertainty in the amount of ambient energy. For applications that require regular power supply, such as mobile services, this uncertainty is not desirable.…”

“…Hence, the harvester characteristics will determine the amount of the harvested energy. Most existing works assumed a linear input-output relationship for the energy harvester [4] - [9]. However, practical harvesters have nonlinear relationships.…”

Outage Analysis and Optimization in Single- and Multiuser Wireless Energy Harvesting Networks

“…57 Under the idealized simplifying assumption of having both 58 noncausal channel-state information (CSI) about the CSI to be 59 encountered in the future and about the EH information (EHI) 60 characterizing the energy arrival rate at the transmitter, in [2] 61 and [3], 1 the optimal offline EUPs were designed for point-to-62 point (P2P) networks using either the throughput maximiza-63 tion or the file-transfer completion-time minimization as the 64 optimization objective function (OF). Later on, the authors in 65 [10] proposed the recursive geometric waterfilling algorithm for 66 solving the same problem, where more efficient recursive com-67 putations were used for finding the optimal solutions. In [4], the 68 authors modeled both the uncertainty of the energy arrival rate 69 and that of the data arrival rate, where the transmission rate to be 70 used was determined by minimizing the average data-buffering 71 delay as the OF.…”

“…T is a solution of 370 (10), which is a high-dimensional system of linear equations. 371 Furthermore, given a certain steady-state probability vector π, 372 it is not possible to derive the buffer-state transition matrix T , 373 and hence, we cannot uniquely and unambiguously determine 374 the discrete EUP L t (l).…”