| Category | Difficulty |
|---|---|
| Quizzes | 7 |
| Presentation | 6 |
| Project | 7 |
This course explores the field of Mobile and Pervasive Computing, also known as IoT (Internet-of-Things) computing. It walks through a number of topics related to the field, including distributed filesystems, virtual machines, wireless technologies, energy harvesting, security and privacy, and much more.
The course is structured as a seminar class, where each week assigns a number of papers for the weekly topic. There is a half-hour quiz at the start of each class, followed by a lecture (either from the professors themselves or guest lecturers) which explores the week's topic in more detail. These papers and lectures are great ways to develop a broad understanding of the field; the papers span from the 1990s to the present, so you'll get to see the history of IoT computing. If you're interested in IoT systems research, or you're looking for papers for a qualifying exam, then this course does a nice job of showing you what's out there.
The second half of each lecture consists of a presentation called a Commercial Scan (or CommScan). For these, one or two students in the class present their own study of the field -- in particular, current commercial practices and challenges faced in the field today. Finally, throughout the entire course, you will work on a project related to IoT computing, either by yourself or with a partner. You'll have a mentor to help guide you, but this is very much an opportunity to apply what you've learned in a hands-on project.
The official prerequisite on the ECE website is 15-410, but you definitely don't need to take that course in order to do well in 18-843. Having some general background in operating systems, distributed systems, and computer architecture should be enough so that the papers are accessible to you. There may be some weeks where the papers are less familiar to you, but that's just fine; that will hopefully provide you with an opportunity to learn something new! Regarding the project, there isn't a defined set of skills that you will need to do well. If you're a senior or graduate student in CS or ECE, then you've likely taken enough courses such that the project will be entirely feasible.
- There are a total of 12 units; there will be an average of 6 papers to read for the following class
- You are expected to go to class, and the quizzes can only be done in class (and not remotely)
- There is no textbook; all papers can be found online
- There is no mid-term nor final exam
- There is a semester long project in pairs (or solo), which is chosen the first day of classes
- There is also a "Commercial Scan" presentation in which you research and present a relevant state of the art in industry, and how commercial practice diverges from academic research
- Project (55%)
- Commercial scan presentation (20%)
- Quizzes (20%) -- you drop the worst test score
- Class participation (5%)
The weekly quizzes are multiple-choice, and they're really intended to check if you've done the readings. Some of the questions can be a bit tricky, but as long as you keep up with the papers, then you should do well on them. The only problem is that the papers themselves can be somewhat lengthy and dense, so you may end up sinking tons of time into 40+ pages of papers each week. One skill that you will develop in the class is how to read papers such that you get the core concepts presented without spending a significant amount of time on them. It's okay if you don't do perfect on the quizzes -- you'll get better at picking out the key ideas as you read more papers!
As mentioned previously, these presentations allow students to explore a topic in greater detail, understanding the challenges that certain areas of IoT computing have faced over the years. The professors provide a list of questions that they want these presentations to address, so be sure to use that to guide your presentation's content. Try to be as clear as possible in what you're presenting, with data and statistics backed by sources and citations. Every student has to do one presentation, but you can sign up for additional presentations; they will take the maximum of your presentation scores, so you might consider doing a second presentation if needed.
The project for this course is more like a capstone, where you have plenty of freedom in how you want to approach it. The professors will have a few projects already available at the start of the class, with mentors for each project who will help provide insight and feedback. You also have the opportunity to bring your own project to the class, though you will need a clearly defined problem statement and an identified mentor for the course.
There are two checkpoints -- one in late September and a second in early November -- which are intended to help you communicate your project's goals and the progress that you've made. Then, there is the final demo in the last week of the semester. Here, you will present your work with a poster, as well as a live demo. Your grade for the project is dependent upon both the checkpoints as well as the final demo and poster.
The project in this course is a unique way to explore a topic of your interest with guidance and feedback from the teaching staff and mentors. For many, this project ends up tying into their research, so this can be an awesome way to make progress on research and get a course credit at the same time!
- The quizzes are closed-book so it's important to read them well to get the general idea, but also take notes since the quizzes tend to ask fine-grained details
- The first day of classes the project mentors (usually Ph.D. students) will present different project topics. At the end of the presentation you walk around the class and find a project you like and also a partner whom to work with (if desired). Each team is working on different projects, meaning no team can work on the same one.
- Some projects are more related to what you learn in class than others. I would pay close attention to the project deliverable requirements since some of them are much more complicated than others.
- I would try to do the commercial scan within the first 3 weeks of classes. Only one commercial scan is required in the semester, and doing it earlier than later can ease the workload when the project deliverables are due.