caidespries.github.io/project3.html at main · CaideSpries/caidespries.github.io · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
<!DOCTYPE html>
<html lang="en">
<head>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>Data Processing Pipeline for a Remote Antarctic Buoy</title>
    <link rel="stylesheet" href="assets/css/projects.css">
</head>
<body>
<div class="container">
    <header>
        <h1>Data Processing Pipeline for a Remote Antarctic Buoy</h1>
    </header>
    <main>
        <section id="project-overview">
            <h2>Project Overview</h2>
            <div style="text-align: center; max-width: 90%; margin: 1rem auto;">
                <p>This project involved designing and implementing a critical data processing system for the SHARC
                    (Sea-ice, Heat, and Atmosphere Research Cabled) buoy, a scientific instrument deployed to study
                    pancake ice in Antarctica. Operating in a remote, power-constrained environment, the system's
                    primary challenge is to collect sensor data from an Inertial Measurement Unit (IMU), process it
                    efficiently, and transmit it securely over an exceptionally costly Iridium satellite link.</p>

                <p>Our solution is an embedded system pipeline built on an ARM-based STM32 microcontroller that performs
                    three key tasks:</p>
            </div>
            <ul>
                <li><strong>Data Compression:</strong> To dramatically reduce satellite transmission costs.</li>
                <li><strong>Data Encryption:</strong> To ensure the confidentiality of the scientific data.</li>
                <li><strong>Data Integrity Check:</strong> To guarantee the data received by researchers is identical to
                    what was collected.
                </li>
            </ul>
        </section>
        <section id="system-architecture">
            <h2>System Architecture and Data Flow</h2>
            <div style="text-align: center; max-width: 90%; margin: 1rem auto;">
                <p>The system is designed as a sequential data pipeline, where raw sensor data is processed through
                    several key submodules before transmission. Each stage must be computationally inexpensive and
                    memory-efficient to operate on the resource-constrained STM32 board. The entire process, from data
                    collection to transmission, is designed to be robust and autonomous.</p>
            </div>

            <figure style="text-align: center; margin: 1rem auto;">
                <img src="assets/img/Projects Docs/project3/system-design-operation.webp"
                     alt="Diagram showing the complete system architecture with IMU sensor, STM32 microcontroller processing pipeline (compression, encryption, integrity check), and Iridium satellite transmission"
                     style="max-width:100%; height:auto;">
                <figcaption>Figure 1: Complete system architecture showing the data flow from IMU sensor through the
                    processing pipeline to Iridium transmission
                </figcaption>
            </figure>
        </section>
        <section id="technology-stack">
            <h2>Technology Stack and Algorithm Selection</h2>
            <div style="text-align: center; max-width: 90%; margin: 1rem auto;">
                <p>Selecting the right technologies was critical to meeting the project's strict power and processing
                    constraints.</p>

                <p><strong>Hardware:</strong> The core of the system is the STM32F051 Discovery Board, featuring a
                    low-power ARM Cortex-M0 processor. For sensor data, we used a SparkFun 9DoF IMU (ICM-20948), which
                    closely mimics the sensor planned for the final buoy.</p>

                <p><strong>Compression Algorithm:</strong> We chose the LZSS algorithm. Our research showed it offered a
                    great balance of high compression ratio, speed, and low computational complexity, making it ideal
                    for this embedded application.</p>

                <p><strong>Encryption Algorithm:</strong> After initial testing, we implemented a custom modified
                    Blowfish algorithm. The popular Twofish algorithm proved too resource-intensive for the STM32's
                    hardware, so we pivoted to its predecessor, which was designed for hardware efficiency and provided
                    robust security.</p>
            </div>

            <figure style="text-align: center; margin: 2rem auto;">
                <img src="assets/img/Projects Docs/project3/SparkFun-9DoF-IMU Breakout.webp"
                     alt="SparkFun 9DoF IMU Breakout board showing the ICM-20948 sensor used for data collection in the project"
                     style="max-width:100%; height:auto; max-height: 400px;">
                <figcaption>Figure 2: The SparkFun 9DoF IMU Breakout board used for collecting motion data</figcaption>
            </figure>
        </section>
        <section id="performance-validation">
            <h2>Performance, Validation, and Results</h2>
            <div style="text-align: center; max-width: 90%; margin: 1rem auto;">
                <p>We conducted extensive testing to validate both the functionality and performance of the entire data
                    pipeline. The system was benchmarked for compression ratio, processing speed, and correctness across
                    various data sizes and conditions.</p>
            </div>

            <div style="text-align: center; max-width: 90%; margin: 1rem auto;">
                <h3>Performance Analysis</h3>
            </div>
            <div class="image-grid"
                 style="display: flex; flex-wrap: wrap; justify-content: center; gap: 1rem; margin-top: 2rem;">
                <figure style="text-align: center; flex: 1; min-width: 45%;">
                    <img src="assets/img/Projects Docs/project3/speed-vs-size-comp-decomp.webp"
                         alt="Graph showing compression and decompression speed versus input size"
                         style="max-width:100%; height:auto;">
                    <figcaption>Figure 5: Compression and decompression performance versus input size</figcaption>
                </figure>

                <figure style="text-align: center; flex: 1; min-width: 45%;">
                    <img src="assets/img/Projects Docs/project3/speed-vs-size-encrypt-decrypt.webp"
                         alt="Graph showing encryption and decryption speed versus input size"
                         style="max-width:100%; height:auto;">
                    <figcaption>Figure 6: Encryption and decryption performance versus input size</figcaption>
                </figure>
            </div>

            <div style="text-align: center; max-width: 90%; margin: 2rem auto;">
                <h3>Complete Pipeline Validation</h3>
                <p>To validate the entire data pipeline, we conducted end-to-end tests that processed sample IMU data
                    through compression, encryption, simulated transmission, decryption, and finally decompression. The
                    final checksum verification confirmed that data integrity was maintained throughout the entire
                    process, as shown in the output below:</p>

                <figure style="text-align: center; margin: 1.5rem auto;">
                    <img src="assets/img/Projects Docs/project3/command-line-output-functionality.webp"
                         alt="Command line output showing full pipeline validation with successful compression, encryption, decryption, decompression, and matching checksums"
                         style="max-width:100%; height:auto;">
                    <figcaption>Figure 7: Complete pipeline validation showing successful data processing with matching
                        checksums
                    </figcaption>
                </figure>
            </div>

            <div style="text-align: center; max-width: 90%; margin: 1rem auto; margin-top: 2rem;">
                <p>Our results demonstrate a highly efficient system. The compression algorithm consistently achieved an
                    average compression ratio of over 15:1, a crucial factor for reducing satellite costs. As shown in
                    the graphs, both the compression and encryption algorithms operate at high speeds (averaging ~950
                    bytes/second and ~910 bytes/second respectively), well within the performance requirements for the
                    buoy's sampling rate.</p>

                <p>We also tested the system's robustness by introducing under-sampling, high-frequency sampling, and
                    Gaussian noise, with the pipeline continuing to perform reliably under all tested conditions.</p>
            </div>
        </section>
        <section id="conclusion">
            <h2>Conclusion</h2>
            <div style="text-align: center; max-width: 90%; margin: 1rem auto;">
                <p>This project resulted in a successful, end-to-end data processing pipeline for a real-world
                    scientific instrument. We met the primary objective of creating a system that could compress and
                    encrypt sensor data efficiently on a low-power microcontroller.</p>

                <p>A key learning experience was the practical challenge of implementation on embedded hardware. Our
                    initial choices of Twofish and LZMA, while theoretically sound, were unsuited for the STM32's
                    architecture. By successfully pivoting to a modified Blowfish and the LZSS algorithm, we were able
                    to meet all user requirements and deliver a robust, working system. The project is a testament to
                    the importance of balancing theoretical performance with the practical constraints of hardware in
                    embedded systems design.</p>
            </div>
        </section>
        <section id="project-document">
    <h2>Project Document</h2>
    <div style="text-align: center; max-width: 90%; margin: 1rem auto;">
        <p>For more detailed information, you can access the following resources:</p>
        <div style="display: flex; justify-content: center; gap: 20px; flex-wrap: wrap; margin-top: 1rem;">
            <a href="assets/docs/Projects Docs/project3/Compression%20and%20Encryption%20Algorithm%20for%20Communications.pdf"
               target="_blank"
               class="button-link"
               style="display: inline-block; padding: 10px 15px; background-color: #3498db; color: white; text-decoration: none; border-radius: 4px;">
                View Project Document (PDF)
            </a>
            <a href="https://github.com/CaideSpries/EEE3097S-Project"
               target="_blank"
               class="button-link"
               style="display: inline-block; padding: 10px 15px; background-color: #333; color: white; text-decoration: none; border-radius: 4px;">
                View Source Code on GitHub
            </a>
        </div>
    </div>
</section>
    </main>
    <footer>
        <a href="index.html">Back to Homepage</a>
    </footer>
</div>
</body>
</html>