Wi-Fi Alliance introduces Wi-Fi WPA3, the next generation of Wi-Fi security, bringing new capabilities to enhance Wi-Fi protections in personal and enterprise networks. Building on the widespread adoption of WPA2 over more than a decade, WPA3 adds new features to simplify Wi-Fi security, enable more robust authentication, and deliver increased cryptographic strength for highly sensitive data markets. As the Wi-Fi industry transitions to WPA3 security, WPA2 devices will continue to interoperate and provide recognized security.
Good news, but it will most likely require you buy a new router, since I doubt many router makers will update their devices to add WPA3 support. I have the last Apple AirPort Extreme, and with Apple exiting the router market, I doubt we’ll see them adding WPA3 support.
…will add support post haste.
Will we finally get the WPA equivalent of HTTPS? Make a secure connection to the router without having to know the password? If not, why do we even bother…
That’s kind of difficult. Encryption isn’t very helpful if you can’t also prove a lack of man-in-the-middle snooping.
HTTPS accomplishes that by using certificate authorities paired with DNS.
WPA currently allows the password to serve a weaker version of that role, since an attacker must first learn the password and can’t discover it by spying on the encrypted traffic.
Without either a certificate authority or some kind of pre-shared key, it’s kind of hard to keep an attacker from setting themselves up for MitM.
Edited 2018-06-27 09:57 UTC
Ok, that makes sense, and of course the WiFi provider could be the MitM themselves, but I’ve always found it weird that in public places WiFi is either unencrypted, or the password hangs on the wall.
You need end-to-end encryption. Anything else is a band-aid.
What WPA is good for (at home) is to prevent your AP from being an unwitting source of attacks/illegal activity.
Enterprise is another matter.
Well, if they’re smart (and they usually aren’t) the wifi network, even the unencrypted one, will have other measures in place. Isolated NAT so that devices can in no way talk to one another to prevent Malware transmission, firewall rules to prevent any bridge from the public wifi to the private LAN, inbound and outbound port blocks as deemed appropriate. Of course, most places that have public wifi haven’t had it audited and wouldn’t employ a network specialist to fix it even if they did.
Tl;dr: no password on public wifi isn’t a problem as long as you take additional security precautions, but most don’t.
Now, if only we can do an encryption between the router and client by default without ever using a password…
Edited 2018-06-27 15:51 UTC
darknexus,
This works for wired connections, but remember that wifi behaves more like a shared medium when it’s signals go through the air. If it’s not encrypted, an attacker can capture/inject packets directly into the air and completely bypass the access point firewalls/nat isolation. There’s nothing an access point can do to protect peers on unencrypted wifi networks.
it’s unfortunate that it’s taken until WPA3 to have encryption by default, since it means it could take another decade for both users and public access point providers to upgrade their hardware. IMHO this should have been addressed years ago in in WPAv1 or even the original WEP.
That was exactly my point
.
ssokolow,
The password method sucks because they aren’t very secure, many places with public wifi have to list the password publicly (because customers want to use it), but obviously it’s vulnerable to rogue APs using the same ESSID + password.
I like the idea of using QR codes to exchange security keys. I assume WPA3 did it right and that the QR code contains some kind of public key rather than just a static password. Another solution could be to use NFC to establish the security.
In principal, an attacker could still perform some physical world attacks by replacing the posted QR codes or NFC links with their own. This clearly leaves evidence, but could still be fairly effective at compromising people. So with this in mind, it could make more sense for us to assume the WiFi is broken and to achieve all security using upstream encryption solutions like HTTPS and VPN.
Edited 2018-06-27 13:54 UTC
As long as there’s a typed fallback. Not everyone is trying to use the WiFi from a phone.
Laptops aside, my OpenPandora (palmtop PC) has no camera or NFC capability and the Pyra (its successor) won’t have them either, as far as I know, even if you order it with the 3G/4G/UMTS option.
ssokolow,
Yeah, we need a technology to bring everything together better. Public keys are too long to type, but the UI could let you download it from the access point itself and then verify the authenticity of a 128bit hash. Incidentally this is what SSH does, but how many people actually take time to verify it versus clicking through?
WPA2 supports a quick connect by pressing a button on the router to keep connections secure once devices are paired. Assuming the initial link is secure, it should remain secure, but there’s a window of vulnerability during the pairing process. It’s hard to solve this problem without introducing new hardware requirements or making the verification step tedious for the user.
Here’s an out of the box idea, more for food for thought that serious, but you can use audio for verification/synchronization as an alternative to QR codes. During pairing, the AP could send out a chirp modulated to contain enough information to verify the connection keys. This way connections to an imposter AP sitting outside in a parking lot would not verify. If the imposter is in the same room, then the sound could alert the user to his presence. Of course at a busy public hotspot, the “noise pollution” could get annoying.
A more practical idea could be to reuse existing certificate authority infrastructure for websites to secure WiFi hotspots as well. For example, to connect to “hotspot.coffeeshop.com”, a hotspot operator who owns that domain could use their SSL certificates to prove that the hotspot is authentic. An attacker could not impersonate the same hotspot since they don’t have the SSL certificate.
Yes WPA3 apparently support opportunistic encryption, aka encryption without password.
But the question is: is it really all that needed? WPA2 is pretty secure and if you choose a very strong password, then I don’t really see where WPA3 excels. Of course, when WPA2 gets deprecated once WPA3 picks up a lot of steam (in 5 years or so?), then WPA3 is the best way to go. But for now, I don’t really see what the benefit over a well-secured WPA2 network.
WPA2 is known broken at this point. See https://www.krackattacks.com/ for info on the specifics. In short, it can force a reset of the replay counter and nonce to the initial values while keeping the same key, which makes it trivial to perform a number of particularly effective attacks on the encryption itself. It’s particularly bad for devices using certain versions of wpa_supplicant for handling the connection setup, as that (had) a bug which would cause a trivially similar attack to trigger installation of a null (all zero) key, functionally rendering WPA2 useless for those devices.
WPA3 fixes these issues.
As a netops person, I’m particularly interested in how long it’s actually going to take for this to roll out to customers. I expect most client devices that actually get updates will have support within a few weeks tops, but just like rolling out a new 802.11 standard, it’s probably going to be far longer before AP support is even remotely widespread.
Hopefully this is enough of a kick in the pants that MS and other client vendors will finally start displaying the exact type of ‘security’ for a wireless connection more prominently (or, you know, at all, Windows 10 still has no quick and easy way to tell if you’re using WEP, WPA1 or WPA2 for an encrypted WiFi connection).