Hi Everyone,
I have an undergraduate degree in aerospace engineering (1993) and soon after graduating became involved in a project where I helped write firmware for a student satellite. I've only worked as an embedded systems engineer since that time--many RTOS and embedded Linux projects usually with an FPGA. Now I'm considering taking time off and getting an online Master's degree with an emphasis in DSP.
I'd like to hear about the different applications members of this forum are designing and how they relate to their studies. Are you doing real-time DSP? Opinions on what has changed over the years and what are the new techniques?
Topics that I've read about over the last year or so that have piqued my interest: very long baseline interferometry, synthetic aperture radar, space communications, and adaptive optics.
Thanks.
Also look at Software Defined Radio. Your FPGA background will be useful for that.
Focus on the basics of signal processing and you can do anything. The math behind the Cooley-Tukey algorithm is based on number theory, and that's fundamental to most of DSP. The math is the key - once that makes sense, all the applications are simple. If you've got the math so it's not confusing (and it will be while learning it!!) then pick the application you like and start there. It should not matter though - it's all just math!
Thanks. Yes I'm aware of SDR.
The reason I'd like to understand the applications and techniques is so that I can evaluate the curriculum of each program. Also if I can get some preliminary understanding from those of you with experience I can use that to have a more meaningful exchange with an advisor going into the program.
No need to quit your job...what did you do between 2 and 4 this morning. Grab Lyon's and other books, see what you can do, do a project for a job demo...It's a tough way to go, but if you are any good, ......
Here is a data point for your survey.
Application areas
Wireless baseband processing
Radar systems / signal design
Geolocation/Geotracking
Communications signal detection and identification
Associated coursework
Statistical communications theory
Digital communications systems
Statistical signal processing
Spectral analysis
Adaptive signal processing
Array processing
Satellite and space communications
Radar signal processing
Kalman/Bayesian stochastic Filtering
Microwave and RF Design
Razed-
DSP is no longer a "destination degree". You can visit, but you don't wanna stay for long. I wrote about this here.
Over the last 10-20 years of your career, at least from an economic perspective, my suggestion would be to focus your MS degree on machine learning, data science, or other AI related area.
-Jeff
Thanks Jeff. I had definitely planned to take at least one course in AI. This is true even though Netflix AI computed "The Godfather II" to be an 87% match for me (it may be my favorite movie).
Regarding servers: I can see, and have indeed worked on, IoT applications whereby real-time hardware is used on a device to process or filter data and then send it to the cloud for further work. For example a series of weather stations that is processing hail strikes and sending features to the cloud to make sense of how the storm is evolving.
So perhaps TI missed an opportunity in a market now dominated by the likes of Nvidia but are you saying that the edge real-time hardware is a goner?
No doubt that embedded system engineering in general -- and edge real-time hardware specifically -- pays well. But it pays better if your MS degree emphasizes AI and your practical experience (and maybe a few courses) include DSP and FPGA.
Let me put it this way -- what do you think your MS advisor is gonna say if you want to do a DSP centric thesis ?
-Jeff
I see. Pay isn't really what driving this. It's more about doing what I perceive as more interesting work by extending my embedded skills. Also I very much doubt I will do a thesis program as those are usually designed for research and that's not a goal.
Interviewing for an embedded software position these days falls into at least two broad categories. One is where the employer is looking more for a computer scientist and the interview will revolve around big O analysis and traditional algorithms. The other is more about hardware and real-time and interrupts etc. along with having decent software design skills (which can be elusive to evaluate). I respect the first camp but prefer the second. I find this to be true at least in the San Francisco Bay Area.
But your comments have been very useful in that I'm trying to understand what courses I should take and AI should be in the mix.
I'm still interested to hear how the different techniques learned through coursework map to actual products and applications. I probably shouldn't have restricted it to DSP.
Its hard to be good at everything but it doesn't hurt to try. I also agree that you don't need to quit working. Just keep knocking the classes down one at a time if need be. The other students are younger but you have years of experience.
As far as real time: feedback controls (like robotics), audio DSP, SDR, radar, some geophysics, medical imaging, etc. Also these days FPGAs are more often being used for DSP. So any project that will help you learn how to pipeline DSP designs would be handy.
After school it is really boils down to what interests you enough to take time to work on it on your own. A lot of I'm working now I learned by doing my own projects.
FWIW,
Mark Napier