We do research on opportunistic and conventional microwave remote sensing from satellite scales to small aerial platforms in environmental sustainability. Our research focuses on recycling the radio spectrum to address the challenges of decreasing radio spectrum space for science while exploring entirely new microwave regions for land remote sensing. To this end, we harness various man-made, non-cooperative Signals of Opportunity (SoOp) which we measure using sensory platforms as widely available as smartphones in addition to dedicated, custom-made instruments to enable a myriad of practical solutions in precision agriculture, forestry, and water conservation. Our research is grouped into six major thrusts:
- – Spectrum Sensing, Coexistence & Recycling
- – EM Modeling & Simulation
- – Information Retrievals & Geospatial Intelligence
- – RF Testbed for Digital Agriculture
- – Off-road Robotics & Mobility
- – Ubiquitous RF Sensing
Spectrum Sensing, Coexistence & Recycling
Our long-term aspiration is to address the challenges of decreasing radio spectrum space for science and to open unusable portions of the spectrum into new methods for science. In particular, we will strive to recycle existing communication and navigation bands to enable a myriad of practical solutions in precision agriculture, forestry, and water conservation. In this thrust, we develop novel implementation of software-defined radio-based receivers from UAS platforms along with a software suite for operating receivers and miniaturized packaging of RF front-end. By opportunistically leveraging the ambient RF communication and navigation signals, these instruments bring us one step closer to maximizing spectrum use for science from small aerial and space platforms. Utilizing such instruments and an advanced simulation tool like our open-sourced SCoBi software will lead to more complex inversion algorithms and higher-quality data products.
EM Modeling & Simulation
In this thrust, we develop advanced EM modeling and simulation tools to (1) explore new measurement techniques and configurations at RF spectrum, (2) appraise the advantages and limitations of each technique, and (3) determine the optimum frequencies and signal properties that can provide geophysical information at high spatiotemporal resolution. Almost any radio signal can be used for opportunistic remote sensing purposes by developing smart processing algorithms backed by physical robust modeling and simulation tools. Our work addresses the following two principal research questions for applications in SoOp research: (1) what radio signals are already available in the surrounding environment? and (2) how can they be recycled for remote sensing? Our open-sourced SoOp Coherent Bistatic (SCoBi) Scattering simulator is in the core of such effort.
Information Retrievals & Geospatial Intelligence
Information retrieval approaches for signals of opportunity must contend with many more unknown variables within the measurement scene and the received signal than traditional microwave remote sensing. The received signal is a composite of coherent and incoherent signals. The received signal emanates from quasi-random locations (i.e., non-repeating ground tracks). A combination of effects from vegetation, topography, surface roughness, soil type, and water bodies under bistatic geometry can suppress the soil contribution. Furthermore, non-geophysical factors such as a variation/uncertainty of transmitter power and the receiver antenna pattern corrections can impact the received signals. Empirical models are great for inversion with limited observations but too simplistic to describe such a complex phenomenon. Sophisticated EM models are excellent for describing the physics but need many parameters that do not exist for each observed pixel for retrievals. While not perfect, machine learning offers some benefits for retrieving non-parametric models. In this thrust, we strive to make machine learning models physically sound and constrained by blending them with physical EM models to improve scientific returns. The best example of this research thrust is the MSU-GRI CYGNSS soil moisture products that are generated from the CYGNSS land observables with the inclusion of multiple remote sensing land geophysical (e.g., topography, land cover, and soil texture) data via the machine learning algorithm.
RF Testbed for Digital Agriculture
Agroecosystems compose large economic sectors in dominantly agriculture-based societies. Microwave remote sensing techniques are the least explored branch of remote sensing for agriculture applications. In this thrust, our purpose is to develop UAS-based RF testbed that can enable implementation of a variety of radio frequency (RF) sensors from UAS platforms and evaluation their use in precision agriculture such as water utilization in agroecosystems. A particular focus is placed on sensing the soil and vegetation traits at multiple depths using signals of opportunity (SoOp). This approach re-purposes satellite communications and navigation transmissions to enable microwave remote sensing at frequencies, fundamentally different than those used in multispectral imaging for UAS-based agricultural mapping and analytics. This can be a powerful augmentation to traditional thermal and multispectral imaging capabilities due to its complementary features such as penetration and their spectrum diversity.
Off-road Robotics & Mobility
Off-road robotics platforms offer a wide range of opportunities for monitoring and exploration via proximal sensing. We can collect in-situ (truth) data along with metadata that could be used to create detailed maps of the measured area. Our research in this thrust focuses on developing unmanned ground vehicle systems with autonomy capabilities for collecting the most useful data about soil and vegetation. In most cases, proximally sensed data provides ground truth for validation and training of machine learning algorithms of aerial platforms. We also envision collaboration and the coordination between aerial and ground robotics system to maximize the efficiency of the joint data collection via a common mission in a more effective and fast way.
Ubiquitous RF Sensing
This thrust originates from the PI’s passion for bringing satellite technology into the hands of ordinary people to foster access to reliable, timely and accurate information to manage their own lands. One of the most effective ways one can monitor environment is through the adoption of satellite technology in our daily lives where it is affordable and usable by everyone everywhere. To bring Signals of opportunity (SoOp) techniques into the mass-market, the PI conjectures that smartphones (without any additional hardware) can be used as GNSS receivers through the use of their internal antennas and GNSS chipsets in order to perform remote sensing of the environment. This thrust explores the possibility of microwave remote sensing using mass-market platforms (smartphones and drones). It is clear that smartphone is not designed for such tasks and needs improvement in hardware or processing approach. The quality of received signals can be improved by combining smartphones with a better external antenna and make other frequency ranges (beyond L-band) possible. We aslo seek use of other integrated wireless chipsets (e.g., cellular carrier’s signal, Wi-Fi, Bluetooth, and FM radio) and external low-cost SDR to extend the use of smartphones in GNSS-R to wider frequency bands to expand the science.
Sponsors
