Performance Characterization of Spatially Random Energy Harvesting Underlay D2D Networks With Transmit Power Control

Abstract

In underlay device-to-device (D2D) networks, the transmitter nodes can harvest energy from downlink cellular (primary) transmissions to solely power the D2D links, which enhances the overall spectral and energy efficiencies. How will the energy harvest and D2D link performance be affected by spatial randomness, temporal correlations, transmit power control, and channel uncertainties? To investigate these issues, we analyze the energy harvesting process of a random (typical) D2D transmitter node, say $\mathcal D_{t}$ , which needs a sufficient harvest to meet the requirements for receiver sensitivity and channel inversion. This system model consists of: 1) three independent homogeneous Poisson point processes; 2) log-distance path loss and Rayleigh fading; and 3) path loss inversion transmit power control. We derive the ambient radio frequency energy at $\mathcal D_{t}$ , and model the harvest as a Gamma random variable. We propose four schemes: 1) single slot harvesting; 2) multislot harvesting; 3) $\mathcal {N}$ slot harvesting; and 4) hybrid harvesting. We develop a Markov chain model for success probability of these schemes, and derive the D2D coverage. We find that a high density of primary transmitters is unfavorable to multislot harvesting for increased D2D link distances. Moreover, hybrid harvesting always outperforms single and $\mathcal {N}$ slot harvesting, and outperforms multislot harvesting except for very high path-loss conditions.