Unmanned Aircraft Systems (UAS), i.e. drones, have been commercially successful in both the consumer and industrial sectors in part due to the wide variety of applications they benefit. In environmental monitoring and precision agriculture, UASs can be utilized for data collection from rural IoT sensor networks. These networks frequently operate over some variant of the IEEE 802.15.4 standard, taking advantage of the standard's low power usage. Consumer 802.15.4 radios are widely available in compact form factors, making them ideal for application in environmental and agricultural sensor networks. Unlike other wireless standards, 802.15.4 is well studied on the ground but has not received rigorous evaluation in three dimensional aerial communication, which introduces new challenges, such as antenna radiation patterns and extreme ranges. This data provides an initial look at the performance of 2.4GHz 802.15.4 for data collection from a UAS. We provide experimental performance measurements using an outdoor aerial testbed, examining how factors, such as antenna orientation, altitude, antenna placement, and obstruction affect signal strength and reception rate of packets.

In March 2019, we performed experimental measurements in an outdoor aerial testbed near our university. In our experiments, IoT nodes running the 802.15.4 protocol were placed on the ground, broadcasting messages at 500ms intervals. The UAS was flown over the network to collect data. We conducted nine total experimental runs divided evenly across three locations. Each run comprised thirteen altitudes, spanning roughly one hour of flight time, for a total of nine hours of flights.

We used six Digi WRL-15126 XBee3 RF 2.4GHz transceivers utilizing 802.15.4. The advertised outdoor range for each node is 1200m at a power of 8dBm and a receiver sensitivity of -103dBm. We selected the XBee3 2.4GHz (as opposed to 900MHz or 868MHz) radio.

Four XBee radios served as IoT transmitters, mounted on a SparkFun XBee Explorer with a USB-to-Serial converter, and controlled by a SparkFun Teensy LC. An external battery supplied power to the Teensy LCs through its own USB-to-Serial converter and forwarded power to the XBee. The nodes were configured to broadcast packets, 23 bytes each, every 500ms. The packet contained unique device and sequence identifiers and a randomly generated floating point number to simulate sensor data.

For each trial, transmitters were placed in a line approximately 11m apart with no obstruction within the 15cm vicinity. XBee placement was randomized for each trial, and GPS coordinates were recorded manually from repeated readings using a smartphone GPS.

The UAS was a DJI Matrice 100, which communicates with a remote control at 5.725 - 5.825 GHz (outside our monitoring frequency of 2.4GHz). We equipped the UAS with two XBees set to receive only. They were mounted to the bottom of the UAS with the \textit{horizontal receiver} parallel to the ground and the \textit{vertical receiver} perpendicular to the ground. Packets and their associated received signal strength indicator (RSSI) reported by the XBee radios were stored via a USB-attached on-board Raspberry Pi2 Model B. The location of the UAS for the duration of each trial was recorded from the Matrice 100 GPS connected via UART to the Pi, sampling at a rate of 50Hz. The UAS was flown in a straight line approximately over the transmitters at an average speed of about 2.2m/s (5mph). The flight path of each trial and altitude varied, as the flights were manually executed under varying wind conditions. Each flight reached a total horizontal distance of 250-300m in both directions from the nearest transmitter (depending on local topography). For each trial, we flew at 13 altitudes (in relation to the lowest transmitter): 30ft, 40ft, 50ft, 60ft, 70ft, 80ft, 90ft, 100ft, 150ft, 200ft, 250ft, 300ft, and 400ft