Member: Giovanna Rilho
Faculty advisors: Narayan B. Mandayam, Ivan P Seskar, Rifat Bin Rashid
LEO satellite mega-constellations (e.g., Starlink, OneWeb, Kuiper) are creating a congested and dynamic spectrum environment, particularly in the 10.7–12.7 GHz band, due to overlapping frequencies and rapid satellite motion. This project will emulate LEO-LEO and LEO-GEO interference in the 12 GHz band using USRPs, incorporating realistic orbital effects (Doppler, path loss, elevation, delay), and measure metrics such as INR, signal degradation, and outages. It will also explore machine learning–based mitigation techniques, including adaptive power control and frequency management, using data from the FR3 testbed.
Week 1 (05/26-05/28)
This week, I began to read previous publications on the topic from WINLAB. I also began Orbit tutorials and built on my knowledge of Linux and Python. Additionally, I identified that the project is studying what happens when multiple satellites use similar frequencies in the 12 GHz band and how a program can reduce that interference.
https://docs.google.com/presentation/d/17-8o3opZuue5OqBvSLMizRQCE3Htvz8H_gwPVBCuJJM/edit?usp=sharing
Week 2 (06/01-06/04)
This week, I began to look into the difference between Low Earth Orbit (LEO) satellites and Geostationary orbit (GEO) satellites. I learned about the architecture of a LEO satellite as well as the orbital effects the satellites experience.
https://docs.google.com/presentation/d/1QiX6tGwv4seeugxlMcLGs6FLNivQAuTEoNe-JoTHfTk/edit?usp=sharing
Week 3 (06/08-06/11)
This week, I began to use the grid to emulate OFDM signals. I set up a transmitter and receiver, then introduced another transmitter to create interference. I altered the center frequency of the interfering transmitter to see how the Bit Error Rate (BER) changed. In another trial, I altered the gain of the interfering transmitter's gain to see how the BER changed.
https://docs.google.com/presentation/d/1w8i8hM7mI7Myf5GAWdGUFi1yfSpB1HQF4ToLvAplbZI/edit?usp=sharing
Week 4 (06/15-06/18)
This week, I began to learn about LEO dynamics and the sources of signal loss that LEO satellites experience. I ran Python scripts that connected to Starlink satellites and conducted handoffs. With these scripts, I was able to see the BER as the signal was transmitted and received.
https://docs.google.com/presentation/d/1muAgETUIHz_k2wbpa9SYIyymHTOSi1nqd0B6izFwyjg/edit?usp=sharing
Week 5 (06/22-06/25)
This week, I used the Pi-Radios to transmit and receive a signal in the FR3 Spectrum band and used the spectrum analyzer to view the signal. I found a paper closely related to this project; however, it is simulation-based and modeled in MATLAB. This week, I worked on recreating their simulation in MATLAB as well as designing the layout for my emulation project on the grid.
https://docs.google.com/presentation/d/1uzt3cNJ7ZrMmFQdfGRHysfZFh2cZMY7KE-QosFJR7x0/edit?usp=sharing
Week 6 (06/29-07/02)
This week, I continued working on the simulation project in MATLAB and produced an INR vs. ITU Threshold graph. For the emulation aspect, I identified the need to create a smart scheduler script and defined all the parameters it must meet. Lastly, this week I started to look into MIMO and digital beamforming.
https://docs.google.com/presentation/d/1LUEJKD087k7wp74GtCMkOyi7rrsQMjkJPThkIPT1AU4/edit?usp=sharing
Week 7 (07/06-07/09)
This week, I got the MATLAB script working that models the paper. For the emulation aspect, I created a Kuiper scheduler script that replicates the Starlink scheduler script and verified that it works and selects satellites. Towards the end of the week, I started to edit the simulation script to match the emulation scheduler. I used TLE Data of the Starlink and Kuiper constellations.
https://docs.google.com/presentation/d/1hTPbRj7Jdv2znQF_M5hY45lacOsf5ZdQ4thG7sTanEw/edit?usp=sharing
Week 8 (07/13-07/16)
This week, I generated synthetic TLE Data for Kuiper Generation 1 deployment based on the 2020 FCC filings and also for the 2024 FCC filings. I compared the inflected interference on Starlink by Kuiper with the current deployment, as well as the full deployment of Generation 1 with the synthetic TLE Data. Proved that there is significant interference with full deployment and can create a protective scheduler for Kuiper to minimize interference on Starlink.
https://docs.google.com/presentation/d/1L54AubmyfJS9WBSd0RjUNyY754r1LLu9IoI-VvZ8g0U/edit?usp=sharing
Week 9 (07/20-07/23)
This week, I created three protective schedulers for Kuiper to utilize and obtained results using the synthetic TLE Data for Gen 1 2024 FCC filings. Starlink uses the same scheduler as previous weeks, and Kuiper runs on one of the three schedulers: separation angle, power control, or separation angle & SNR floor. Seperation angle & SNR floor showed the best overall results measuring percentage of time crossing ITU threshold value and SINR for both systems.