Microwave ovens, wi-fi, and HTTP
Have you ever noticed a correlation between a running microwave and your internet connection stuttering?
In my childhood home, I have a Pavlovian response to the sound of someone opening the microwave; deep down, I know that I'm getting disconnected from the wi-fi. My wife sighs when I tell her our video call might disconnect. Back in high school, my gaming friends called it “Bobbie got ganked by microwave”: it was like getting ambushed by the enemy team, except more embarrassing. Just the electromagnetic interference of a bowl of soup being heated up.
This is no coincidence. Wi-fi can operate in the 2.4 GHz frequency range, which penetrates walls and can carry a decent amount of data. Microwaves generate electromagnetic noise in that same 2.4 GHz frequency range - this same frequency causes water (and other materials in your food) to vibrate and heat up. That microwave noise obliterates the structured data in the WiFi signal, like people talking over each other in a crowded room.
We can thank the International Telecommunication Union (ITU) for this collision.
The ITU and the ISM bands
The ITU coordinates international standards for telecommunications. Assigning the usage of different frequencies helps users avoid interfering with each other. In 1947, multiple countries asked the ITU to carve out certain frequency bands for non-communication uses. The 1947 proposal from the United States mentions several valuable non-communication use cases, including:
Industrial, like “the rapid drying of glued joints and plywood” or “controlled surface hardening of metals”
Scientific, like “the lab testing of new materials and equipment”
Medical, like uses in “surgical operations” or inducing “artificial fever”
The microwave oven itself is explicitly mentioned in the proposal:
“A recent important development of industrial apparatus is an electronic cooker which not only cooks food evenly from the inside out, but completes the cooking process in a matter of seconds as compared to minutes with older methods.”
From the ITU Archives and Library Service: Documents of the International Radio Conference (Atlantic City, 1947) (page 464), which I found via Adam Haun’s question on Electrical Engineering StackExchange
As a result, the ITU created the Industrial, Scientific, and Medical (ISM) frequency bands, to support these newer use cases. The ITU's international recommendations were adopted by national organizations, including the Federal Communications Commission (FCC) in the United States.
The FCC and spread spectrum modulation
Over the next decades, “spread spectrum modulation” emerged as a new technique in military and law enforcement communications: by spreading the data across a wider range of frequencies, it makes a particular data transmission both harder to eavesdrop on, and harder to interfere with or "jam". Prior to 1985, the FCC did not authorize any usage of spread spectrum for civil use cases. But in 1985, the FCC removed restrictions on spread spectrum because it had “several interesting civilian applications":
“The Commission proposes to accommodate spread spectrum systems by reducing regulation to the maximum extent feasible. The Commission believes that such action will lead to a more rapid development of spread spectrum technology in the civilian sector.”
From Gen. Docket 81-413; FCC 81-413, via IEEE802.org
These new civilian use cases (including cordless telephones) were allowed to use the ISM bands without any licensing. These new communication use cases had strict power limits, weren't supposed to interfere with existing ISM use cases, and “must accept any interference which these systems may cause to their own operations.”
Modern wi-fi, as well as technologies like Bluetooth and proprietary protocols like the one Logitech uses for wireless mice and keyboards, descend from these spread-spectrum use cases on the 2.4 GHz ISM band.
So, when I complain about the microwave interfering with my wi-fi - well, the microwave was here first, and my wi-fi router does accept the interference.
The “wild west” of the ISM bands, and how they relate to HTTP
These days, most wi-fi devices support the 5 GHz band, a different frequency band that does not compete with the microwave and other ISM uses (and despite the similar naming, completely unrelated to 5G cellular, where the “G” stands for generation of cellular technologies rather than frequency). After the technology was proven out, industry working groups were able to standardize new versions of wi-fi that used different frequency bands, avoiding the issues of the crowded 2.4 GHz space.
I think there are analogues to HTTP, the very HyperText Transfer Protocol that delivers this webpage.
HTTP (and the World Wide Web) was not designed or decreed to be a standard (unlike the “narrow waist” of the Internet Protocol (IP)).
Image by Steve Deering, via “The Internet Was Designed With a Narrow Waist” (oilshell.org)
HTTP was originally just another application protocol, competing against protocols like Gopher. Eventually, HTTP won out and its success meant that everyone supported HTTP (and HTTPS).
It's easy to write new protocols on top of HTTP. You don't need to convince a corporate IT team to open a new firewall. You don't need to convince a networking hardware provider to not drop your packets. You certainly don’t need to convince the FCC to dedicate some space for your new protocol. By default, everyone has reasonable access to HTTP.
When the need for your new protocol is justified and you're running into the limitations of HTTP, then you can create a new protocol, with the momentum of your existing users and use cases. For example, Websockets descended from HTTP long polling; eventually creating the entirely new Websockets protocol helped overcome scaling and performance limitations baked into HTTP.
(Side note: I’m definitely not the first to notice this - this report from Lucian Popa and Patrick Wendell and Ali Ghodsi and Ion Stoica of UC Berkeley EECS said that “HTTP is poised to become the de-facto ‘narrow waist’ of the modern Internet” back in 2012 (and as a double side note, those latter three authors are likely better known as Databricks co-founders). And RFC 9205, first drafted in 2017, explores similar ideas.)
It's useful to have a wide-open place for experimentation. Without somewhere to try new things, it will be difficult to explore the design space and learn. New technology will emerge from the places where they are unconstrained.
And the next time your internet suddenly drops while someone’s heating up leftovers - well, now you know why. But first - is your microwave actually running? You’d better go catch it then!