摘要： This chapter considers a point-to-point EH wireless channel when the transmitter and the receiver are powered by EH. Under this setup, we first maximize various system utilities (such as the end-to-end throughput and the non-outage probability) via adaptive power allocation over time subject to the EH constraints at the transmitter, where the accumulatively consumed energy should not exceed that accumulatively harvested at any time. Under various CSIT and ESIT assumptions, we discuss the respective optimal designs. For example, in the noncausal CSIT and ESIT case, the optimal offline power allocation is shown to follow a non-decreasing and piecewise-constant (or staircase-like) pattern for the Gaussian channel case, and the optimal transmit power corresponds to a staircase water-filling power allocation for the fading channel case. In the causal ESIT and CSIT case, applicable online optimization strategies include dynamic programming-based and heuristically designed power allocation strategies. We also point out several open problems under different ESIT and CSIT considerations. For example, the throughput maximization problems under the case with causal CSIT and noncausal ESIT and the case with no CSIT and noncausal/causal ESIT remain unsolved in general, while the outage probability minimization problems under causal CSIT and noncausal/causal ESIT are also open problems at large. Furthermore, we extend the power allocation policies to other scenarios by considering other practical constraints such as limited battery capacity and imperfect circuits. We show that in the noncausal CSIT and ESIT case, the staircase power allocation with Gaussian channel and the staircase water-filling power allocation with fading channels can be modified to take these new practical constraints into account. With imperfect circuits, the EH transmitter should employ an on-off power allocation, which turns the transmitter into a sleep mode to save power in order to transmit more energy efficiently in the wake-up mode to tradeoff between the transmit energy consumption versus the circuit energy consumption. These new offline power allocation strategies motivate new online power allocation approaches with only causal CSIT and ESIT. Finally, we provide discussions on the transmit and receive power allocations in the case with EH transmitter and EH receiver, where in addition to the transmitter, the receiver's power consumption is also subject to an EH constraint. Under such a scenario, it is essential for the transmitter and the receiver to jointly optimize the transmit power and the code rate (hence, control the receive power consumption) to improve the end-to-end utility. Note that the case of EH receiver has not been widely investigated in the literature and there are rich open problems (e.g. under causal CSIT and ESIT) worth being pursued in future work.