Throughput Maximization of Wireless-Powered Communication Networks An Energy Threshold Approach
Abstract : In this paper, we consider a wireless-powered communication network (WPCN) where one mobile hybrid access point (HAP) coordinates the wireless energy transfer to sensor nodes and receives data from sensor nodes, which are powered exclusively by the harvested wireless energy. As the harvest-then-transmit protocol is employed by sensor nodes, a major challenge lies on the tradeoff between achievable throughput and energy harvesting opportunity of sensor nodes. Confronting this challenge, we develop an energy threshold approach by jointly considering geographic locations and energy states of sensor nodes, where wireless energy transfer occurs when none of the sensor nodes in the range of data transmission has more energy than the threshold, otherwise data transmission from one randomly chosen qualified sensor node to the HAP occurs. By comparing the range of energy harvesting and that of data transmission, we divide the network topology into two cases for throughput analysis, and formulate the energy states of sensor nodes as Markov chain processes with different energy state spaces in the two cases. Through monotonicity analysis of achievable throughput and probability distribution of energy states, we prove the existence of the optimal energy threshold that maximizes the achievable throughput, and find that the achievable throughput under infinite battery size could be viewed as the upper bound of that under the limited battery size. Finally, simulation results validate theoretical results of the optimal energy threshold, and show the impacts of system parameters on the achievable throughput.
? Deep learning-based algorithms have great potential in wireless powered communication networks, providing competitive performance with less complexity in comparison with existing solutions.
? They provided a fairly accurate approximation of a popular algorithm, named weighted minimum mean squared error (WMMSE).
? The DNN architecture accepts the channel coefficient as input and gives an optimized solution as output.
? After harvesting energy from the H-AP, the kth user sends an information signal utilizing the energy harvested in the downlink phase.
? To maximize the sum throughput over a finite horizon, the initial optimization problem is separated into two sub-problems and finally can be formulated into a standard box-constrained optimization problem, which can be solved efficiently.
? Wireless power transfer (WPT) using radio frequency signals is attracting attention as a viable approach to the energy harvesting problem.
? We consider the problem of maximizing the sum throughput over a finite horizon with energy saving.
? Prolonging the lifetime of battery powered devices in wireless networks is an important problem.
• The proposed deep neural network accepts the channel coefficient as an input and outputs minimized power for this channel in the WPCN.
• The proposed approach ensures the quality of service (QoS) of the WPCN by managing user throughput and by keeping harvested energy levels above a defined threshold.
• Many researchers have already proposed solutions that use numerical optimization to solve signal processing tasks.
• They proposed a deep autoencoder as a solution to learn the channel parameters autonomously at the energy transmitter based on the feedback from the energy receivers.
? To obtain the optimal solution, the initial optimization problem is separated into two sub-problems and finally is formulated into a standard box-constrained optimization problem, which can be solved efficiently by the trust-region-reflective algorithm.
? WPT systems can simultaneously convey energy and information on the wireless signals and the inherent tradeoff between information rate and power transfer efficiency has been recently characterized.
? The former is a convex optimization problem, which gives us a closed-form relation between the time allocation of donwlink WET and uplink WIT and the latter can be formulated as a standard box-constrained nonlinear programming problem, which can be solved efficiently using the trust-region-reflective algorithm.
|