Everything I've ever seen with a 2GHz radio supports b, [0] and with a 5GHz radio supports a.
[0] Even though some AP admins deactivate 802.11b support for entirely valid throughput raisins. AIUI, 802.11a doesn't suffer from those issues.
Example: I created wifi drivers for a handheld platform back in the 90's (Magic Cap), including support for the earliest 802.11 PC cards. Fans of the platform (both of them) encouraged me to update the drivers so they could continue using their handhelds over wireless, but it's simply not possible to get a 1999 wifi card on a modern wifi network in any half-secure fashion.
True. It's a damn shame about WEP, and WPA2 didn't happen until 2004. But, 2004 is eleven years ago. Disgregarding the shitfest that is WPS-PIN, WPA2 with CCMP is looking rather good.
Are you aware of breaks in WPA2 with CCMP-only crypto and either a well-chosen passphrase, or WPA2-Enterprise with a proper client authorization scheme?
> ...it's simply not possible to get a 1999 wifi card on a modern wifi network in any half-secure fashion.
Not saying that there are any WiFi cards that behave in this manner, [0] but would a hypothetical WiFi card that only knows how to how to establish and pass around -er- frames(?) on a 802.11a/g link and relies on the client PC to use the frames passed back and forth to do link crypto & etc be
* Possible
* Reasonably sufficiently fast, given enough CPU power on the client
* Able to cope with any likely enhancements (such as the addition of WPA2 back in 2004) with a driver update, rather than a firmware update
?
[0] Even though I thought that pre 802.11ac Atheros cards had open firmware and open drivers, so one got pretty close to this.
Ethernet via RJ45 has really good staying power; IEEE standard in 1990. A refrigerator at 10Mbps is probably sufficient, but if not, 10GBase-T is approaching affordability, and a standard for 40GBase-T is expected next year, although you may need to pull new wires.