UAV-to-UAV mmWave Channel Model: paper and dataset

This webpage describes the main results and the dataset associated to our paper M. Polese, L. Bertizzolo, L. Bonati, A. Gosain, T. Melodia, An Experimental mmWave Channel Model for UAV-to-UAV Communications, in Proc. of ACM Workshop on Millimeter-Wave Networks and Sensing Systems (mmNets), London, UK, Sept. 2020.

• Paper link: arXiv.org pre-print
• Dataset and code: github.com repository
• Video of the experiments: youtube.com

Unmanned Aerial Vehicle (UAV) networks can provide a resilient communication infrastructure to enhance terrestrial networks in case of traffic spikes or disaster scenarios. However, to be able to do so, they need to be based on high-bandwidth wireless technologies for both radio access and backhaul. With this respect, the millimeter wave (mmWave) spectrum represents an enticing solution, since it provides large chunks of untapped spectrum that can enable ultra-high data-rates for aerial platforms. Aerial mmWave channels, however, experience characteristics that are significantly different from terrestrial deployments in the same frequency bands. As of today, mmWave aerial channels have not been extensively studied and modeled. Specifically, the combination of UAV micro-mobility (because of imprecisions in the control loop, and external factors including wind) and the highly directional mmWave transmissions require ad hoc models to accurately capture the performance of UAV deployments. To fill this gap,

• we propose an empirical propagation loss model for UAV-to-UAV communications at 60 GHz;
• we release the dataset of aerial measurement campaign conducted with the Facebook Terragraph channel sounders mounted on two DJI M600;
• we compare the channel model with state-of-the-art path loss models (e.g., 3GPP).

UAV-to-UAV mmWave Channel Model: dataset and code structure

The file dataset.csv contains the measured data as processed by Facebook Terragraph sounders.

The MATLAB scripts in the Github repository can be used to load and manipulate the dataset, and to generate the results of the paper. You can load and save the processed data in a .mat file using load_dataset.m. Then, the scripts fit_6.m, fit_12.m, fit_15.m and fit_6_12_15.m perform the fit on the measurements. Finally, compare_height.m, compare_3gpp.m, compare_beam_dist.m, and compare_beam_error_deg.m generate the plots shown in the paper.

The parameters in the database.csv are defined as follows:

• distance: Distance between the UAVs

• altitude: Altitude of the UAVs from the ground

• link_dir: Direction of transmission, e.g., "1to2" indicates that node 1 is transmitting and node 2 is receiving

• tx_beam, rx_beam: Transmitter and receiver beam indices used for scanning

• mcs: IEEE 802.11ad Modulation and Coding Scheme (MCS)

• tx_mask*, rx_mask*: Transmitter and receiver antenna masks. Defines which antenna elements are active. The three hexadecimal numbers in the mask represent the active state of the specific tile (*_mask22, *_mask23, *_mask24), where a 1 in a binary position represents an active and a 0 represents an inactive element. E.g., "0xfff" specifies that all beams are active on the specific tile.

• tx_gain_idx: Transmitter gain indices

• rx_rf_gain_idx, rx_if_gain_idx: Receiver gains indices from the Automatic Gain Control (AGC) closed-loop feedback regulating circuit in the node amplifier chains

• stf_snr: SNR measure based on spatiotemporal filtering (STF)

• post_snr: Post-equalization SNR

• rssi: Received Signal Strength Indicator (RSSI)

• rssi std: RSSI standard deviation

• tx temp, rx temp: Transmitter and receiver nodes junsction temperature

• eirp: Effective Radiated Power (EIRP) at the receiver node

• pRx: Received power at the receiver node

• path_loss: Signal path loss calculated from the difference of EIRP and received power