Nearly 9 months ago I left the working world behind to embark on my PhD. I’m lucky to have this opportunity and think it’s worthwhile to log the journey. I intend on this becoming a bi-weekly (twice monthly) occurrence, where I’ll recount my progress (if there is any) and consider what’s coming next.
My field is Computer Science, I work in the intersection of wireless communications, computer networking and machine learning. I spent a few years in software engineering and studied A.I previously, but with such a broad territory to cover I still had a lot to learn at the outset. I had no grounding in the wireless / networking side of things so for the first 3 months I gave myself a rough education in these topics (reading papers, textbooks etc).
After this period I was eager to jump in to ‘real’ work. I had a budding interest in cognitive radios and spectrum sharing, so I began working on a radio-channel selection algorithm. This work didn’t amount to much in the end but it ultimately spurred me on in a more fruitful direction. I spent the first half of November clobbering together a paper for ICC, titled ‘Recurrent Neural Networks for Proactive Channel Switching in Intelligent WiFi access points’. It’s a mouthful, but the underlying idea is relatively simple.
– Recurrent Neural Networks (RNNs) are good at predicting sequences of data.
– Radio Signal Strength Indicator (RSSI) values tell us about interference levels.
Take the two points above, jam ’em together, and you get the basis for a paper (whether it’s a good one or not remains to be seen..); RNNs can predict future RSSI sequences and inform radio-channel switching policies for interference avoidance.
That’s it so far, there’s a bit more to it (including some high performance computing courses), but that’s the gist. 09/01/2023