4 ms·
This seems like the right approach--airgapped with redundancy. Not only is it more dependable and secure, it's also cheaper. You already have a smoke detector,
by hellcow 11y ago
This seems like the right approach--airgapped with redundancy. Not only is it more dependable and secure, it's also cheaper. You already have a smoke detector, so keep the parts that work.
That raises an interesting question: what other "dumb" devices can we connect to the internet safely through that same method?
- nickpsecurity 11y agoNote that my RF method is extremely simple, dirt cheap, and has decades of field use. All kinds of industrial plants used that trick for local monitoring systems. Plus, it's strictly one-way transmission: can't hack it. Devices on receiving end are simple enough to be made as secure (or not) as you'd like.
- mindslight 11y agoAnything that puts out simplistic PDUs, ideally one way only. A smoke alarm sends one one-way one-bit message, which can be done with a broadcast in any medium. You could even achieve similar isolation even with an optoisolated RS-232 port if the protocol is auditable, non-general, and non-extensible. But good luck actually guaranteeing those things! Internet access is akin to Turing completeness for programming languages - seemingly beneficial, easy to accidentally obtain, and ultimately a curse.
- nickpsecurity 11y agoYour profile statement is interesting given Im going opposite route: from high-assurance security in software to learning hardware cuz software cant free us on subverted, inherently-insecure hardware. Gotta tweak tge hardware a la crash-safe.org or Cambridge's CHERI cpu to make job tractablw. Im still looking for EE's to run my tooling ideas and risk assessments by but those that help are uncommon. Send me an email if your interested in talking secure HW/SW architecture or methods.
- ashitlerferad 11y agoTalk to the lowRISC folks?
- nickpsecurity 11y agoThat's on my TODO.