Modeling Realistic Bit Rates of D2D Communications between Android Devices

Although D2D communications have been extensively investigated in the literature, relatively few works have focused on understanding the capacity of direct links in a real setup. In this paper, we propose an empirical characterization of the currently available high-speed D2D technologies in Android, namelyWi-Fi P2P andGoogle Nearby. To this end, we developed a custom Android application calledOcat which interacts with the available D2D APIs and measures the link's goodput. From the experimental campaign, we derive several useful observations. Concerning communication capacity, the goodput between Android devices ranges between 320~Mbits/s when nodes are within 20 meters of each other and 0.1~Mbits/s when the distance grows to 300 meters. Based on the experimental measurements, we propose a model of the upper-bound goodput as a function of the distance between two devices. Using the wireless signal strength as a link measurement, we combine it with the two-ray ground-reflection model to infer the goodput and obtain a good fit for the characterization of D2D links between Android devices. Our findings provide a reality check in regards to actual direct data-exchange capabilities of Android devices and can help assess system performance of D2D applications.