I'm quite interested in 802.11 over amateur radio frequencies; however, all I have done at this point is reading (I don't have a ham license yet). The conventional wisdom (espoused in this ham.SE answer) is that you cannot encrypt radio communications at all. This limitation seems to fly in the face of normal countermeasures used to secure data transfers through the internet.
So encryption is off the table; however, that still leaves issues such as authentication and data integrity. I have read that CRAM-MD5 hashes are sometimes used as part of data services implemented over 802.11 ham radio, because CRAM-MD5 hashes are allowed under the FCC rules. The contents of messages signed with CRAM-MD5 are always visible; yet, the authorship of the message can be verified (because CRAM-MD5 works in the same way that PGP does). As best I can tell, using CRAM-MD5 leaves you vulnerable because:
- It does not perform mutual authentication (i.e. the client can't verify the server's identity)
- It's possible to run off-line dictionary attacks against captured hashes (although the underlying HMAC-MD5 hash doesn't have the same vulnerabilities as MD5).
The bigger concern to me is a man in the middle attack, since CRAM-MD5 offers no mutual authentication between client and server. This kind of vulnerability is also the kind of dynamic that makes PEAP worse than no 802.11 authentication (since PEAP doesn't enforce mutual authentication between the wifi client and RADIUS server, like EAP-TLS does). It has been shown that weak PEAP clients freely offer their password hashes to a wifi AP masquerading the same SSID as the real AP. My concern with 802.11 over ham frequencies is determining that I'm giving my password / data to the right endpoint on the other side (and not an attacker).