Throughput maximization for energy harvesting nodes transmitting over time-varying channels

We investigate in this work the throughput maximizing transmission strategies of an energy harvesting transmitter which communicates over a time-varying channel. Two system models are considered to this end, the generic model for which the energy harvesting node is treated as a continuous-time control system, and the MQAM model for which practical modulation schemes are employed for transmission. A generalized circuit power function dependent on the transmit power is formulated for the generic model. We apply optimal control theory with this model and propose a modified water-filling algorithm to obtain the optimal transmission policy. The circuit power modelling with MQAM transmissions, on the other hand, is based on studies on the power dissipation of various functional components of the transmitter circuitry. Energy-efficient modulation orders can be found and the throughput maximization becomes a linear optimization problem in this case. A heuristic algorithm with near-optimal performance is also developed to conquer the difficulty of problems with large number of variables. Besides, the dynamic programming (DP) technique is applied with both system models which not only verifies the optimality of our obtained power allocations, but also provides an online solution to the situation that the transmitter does not have in advance the full channel state information (CSI) of the entire time slot of interest.