2019/08/30 by Morteza Varasteh, Jakob Hoydis, Varasteh, Morteza +3 · 1 citation
Engineering · #Energy Harvesting in Wireless Networks #Wireless Power Transfer Systems #Innovative Energy Harvesting Technologies
paper · pdf · doi:10.48550/arxiv.1908.11726
Nonlinear energy harvesters (EH) behave differently depending on the range of\ntheir input power. In the literature, different models have been proposed\nmainly for relatively small and large input power regimes of an EH. Due to the\ncomplexity of the proposed nonlinear models, obtaining analytical optimal or\nwell performing signal designs have been extremely challenging. Relying on the\nproposed models in the literature, the learning problem of modulation design\nfor simultaneous wireless information-power transfer (SWIPT) over a\npoint-to-point link is studied. Joint optimization of the transmitter and the\nreceiver is implemented using neural network (NN)-based autoencoders. The\nresults reveal that for relatively small channel input powers, as the power\ndemand increases at the receiver, one of the symbols is shot away from the\norigin while the remaining symbols approach zero amplitude. In the very extreme\ncase of merely receiver power demand, the modulations are in the form of On-Off\nkeying signalling with a low probability of the On signal. On the other side,\nfor relatively large channel input powers, it is observed that as the receiver\npower demand increases, a number of symbols approach zero amplitude, whereas\nthe others (more than one symbol) get equally high amplitudes but with\ndifferent phases. In the extreme scenario of merely receiver power demand, the\nmodulation resembles multiple On-Off keying signalling with different phases.\n