Preksha H Patwa

GE4050

PGE22258

In response to the pressing need for innovative approaches to securing maritime and national borders, this project spearheads a revolutionary shift in surveillance capabilities by developing an advanced in-orbit data pre-processing system tailored for optical, multispectral, and synthetic aperture radar (SAR) Earth observation (EO) imagery. Leveraging the extensive capabilities of EO satellites, the initiative aims to enhance operational efficiency and reduce latency in monitoring critical zones. By transitioning from conventional ground-based operations to in-orbit processing, the project promises heightened efficiency and reduced data transmission latency, marking a significant milestone in Satellite Edge Computing. The core focus lies in crafting a sophisticated in-orbit data pre-processing pipeline, which involves refining data from diverse satellite-based sensors and directly transforming Level 0 (L0) to Level 2 (L2) data on satellite edge-based platforms. To infuse computational sophistication, Graphics Processing Units (GPU) and Field-Programmable Gate Arrays (FPGA) are integrated onboard. The project progresses through stages of assessing requirements, defining data architecture, crafting the pre-processing pipeline, and conducting critical evaluations before engaging in machine learning (ML) feature extraction. Collaborative efforts with stakeholders are pivotal in integrating necessary corrections and refinements. The project transitions into the build and test stage, deploying and rigorously testing the data pre-processing pipeline on satellite data processing units (DPU), with an iterative approach emphasizing deployment, testing, and reiteration based on feedback and identified areas for improvement. Comprehensive testing and reporting follow, with the system undergoing rigorous evaluation on three distinct DPU platforms. The findings are synthesized into a detailed report, documenting the testing process and providing valuable recommendations for future satellite missions dedicated to enhancing maritime and border security, thus representing a significant stride towards real-time surveillance capabilities and offering a blueprint for future missions in the domain.


Report Content

Topic of DRP

Aim and Objective

Introduction

Image Preprocessing

Data sets

Methodology

Cloud detection

Conclusion

Application

References