GP00 - Edge Vision (Grad Students Only)

Project Overview

Each graduate student will independently select a unique, practical edge computing application that could realistically be implemented using today’s technology (or technology expected within the next ~5 years). Examples include sports/entertainment, manufacturing, environmental monitoring, healthcare, transportation, disaster response, smart cities, etc.

No coding, no prototypes, no working demos are required.

The focus is on theoretical analysis, architecture design, and reasoned estimates grounded in concepts taught in class.

Deliverables

• 15-minute in-class presentation (with slides)
• 5-page report (~1000 words)

Suggested Project Ideas Vignettes (to spark your thinking)

These are short examples, choose your own topic (it must be different from your classmates’ we will have sign up sheet).

1. Stadium AR Overlays
   Real-time player tracking and statistics delivered via fan mobile apps using edge servers in/near the stadium, achieving ~5 ms latency for smooth augmented reality overlays.

2. Factory Predictive Maintenance
   Vibration, temperature, and acoustic sensors on industrial machines process data locally at the edge to detect anomalies and predict failures before they cause downtime.

3. Urban Air Quality Monitoring
   Distributed low-power IoT sensors across a city aggregate and analyze pollution data at nearby edge nodes, enabling instant public health alerts without round-trip cloud delays.

4. AR-Guided Surgery
   Hybrid edge-cloud system in a hospital network provides sub-10 ms latency AR overlays for surgeons (e.g., highlighting vessels or tumor margins) using local 5G MEC infrastructure.

5. Smart Intersection Traffic Optimization
   Cameras and sensors at major intersections use on-site edge computing to dynamically adjust traffic light timing, reducing congestion and improving flow in real time.

Required Content (must appear in both presentation and report)

1. Motivation and Use Case
   • Paint a vivid, compelling picture of the problem and how edge computing dramatically improves the outcome.
   • Describe real-world benefits (lower latency, higher reliability, bandwidth savings, privacy, etc.).
2. Key Class Concepts
   Explicitly reference and apply Edge computing concepts:
   • latency reduction
   • task partitioning / offloading decisions
   • hybrid edge–cloud processing
   • resource allocation and orchestration
   • networking approaches
   • bandwidth/latency trade-offs
3. Architecture Diagram
   Include at least one clear diagram showing:
   • data sources (sensors, cameras, devices)
   • edge nodes / servers
   • cloud connection (if any)
   • flow of data and processing decisions
4. Processing Details and Estimates
   Theorize a realistic processing split:
   • What % of computation happens at the edge versus cloud?
   • What latency targets are you aiming for (e.g., <5 ms for real-time tasks)?
   • What kinds of workloads run where? (inference, filtering, aggregation, etc.)
   • Rough estimates of required edge compute (cores, memory, throughput), data rates, etc.
5. Trade-offs and Conclusion
   Discuss advantages of your edge approach versus pure cloud.
   Mention limitations, scalability considerations, and future potential.

Include proper citations (class slides/notes + 3–5 external sources, e.g., ACM or IEEE papers, industry reports).

Project Timeline

1. Topic Selection Due: February 10 by 11:59 PM

• Select a project topic
• Sign up using the Topic Sign-Up Sheet

2. Project Abstract Due: March 9 at 11:59 PM

• Upload the abstract PDF here, with filename: <firstname><Lastname>Abstract.pdf.
• The abstract should outline the project scope and content to be covered in the final report and presentation.

3. Final Presentation and Report Due:April 6 at 11:59 PM

• Upload the presentation PDF here, with filename: <firstname><Lastname>Presentation.pdf.
• Upload the written report PDF here, with filename: <firstname><Lastname>Report.pdf.
• In-class presentations will be scheduled later in the semester.

Presentation Timing (15 minutes total)

• Motivation and Use Case → 5 minutes (make this engaging, tell a story)
• Architecture and Class Concepts → 4 minutes
• Processing Split and Estimates → 3 minutes
• Trade-offs, Scalability and Conclusion → 3 minutes

Practice to stay within time. Use clear slides with diagrams, avoid text-heavy slides, and DO NOT go overtime.

Report Structure (5 pages, ~1000 words)

1. Title page (not counted in 5 pages)
2. Motivation and Use Case (~1 page)
3. Architecture and Class Concepts (~1.5 pages, must include diagram(s))
4. Processing Details and Estimates (~1 page)
5. Trade-offs, Conclusion and References (~1 page)

Format: 11 point font, 1-inch margins, single spacing. Include 2–3 figures/diagrams.

Evaluation Criteria

• Clarity and storytelling (30%)
• Depth of technical analysis and use of class concepts (40%)
• Quality of visuals, diagrams and citations (20%)
• Adherence to time/page limits and professional presentation (10%)

Choose a topic that excites you, then analyze it deeply and clearly. Good luck!