802.1X and MAC Authentication Bypass
802.1X is the IEEE standard for port-level access control: until a device authenticates, the switch port forwards no user traffic. MAB is the fallback for dumb devices that cannot run an 802.1X client, and understanding the gap between the two matters when designing an access policy.
The three roles in 802.1X
Supplicant. The client software running on the device that wants to connect.
Authenticator. The switch or access point. It makes no decision itself; it relays authentication messages and opens or closes the port based on the result.
Authentication server. Usually a RADIUS server, and the party that actually validates identity.
The flow runs roughly like this. On connection the port is unauthorised and passes only EAPOL frames. The switch prompts the device, the device submits credentials, and the switch wraps them in RADIUS and forwards them to the server. The server returns accept or reject, the switch opens the port accordingly, and it can push down VLAN and ACL policy at the same time.
The essential point is that the credential is independent of the device. A username and password, or a certificate, belongs to a person or a machine identity, not to a MAC address. That is what makes the authentication auditable and revocable.
How MAB works
Printers, IP cameras, door controllers and older IP phones frequently have no 802.1X client. Requiring authentication would leave them permanently offline.
MAB works like this: after 802.1X authentication times out on the port, the switch submits the device's MAC address to the RADIUS server as both username and password. The server checks a pre-maintained MAC whitelist and permits the device on a match.
Configuration usually tries 802.1X first and falls back:
interface GigabitEthernet1/0/10
authentication port-control auto
authentication order dot1x mab
authentication priority dot1x mab
mab
dot1x pae authenticator
MAB offers very limited security
It has to be clearly understood that the MAB credential is the MAC address, and a MAC address travels in cleartext and can be set to anything.
An attacker only needs to sniff one whitelisted address and set their adapter to that value to pass authentication. Doing so requires no skill at all, as covered in changing a MAC address on each platform. It is the same reasoning as why MAC filtering is not a security measure.
So MAB should be understood as an operational exception path rather than a security control. Sensible practice looks like this:
Keep the scope of MAB as small as possible. Enable it only on ports that genuinely need it, never globally.
Give MAB devices a restricted VLAN and ACL. Passing authentication should not grant full internal access, only the destinations and ports the device actually needs. A camera that talks to a recording server has no business reaching finance systems.
Combine it with device profiling. Beyond checking the MAC, the RADIUS server can examine DHCP fingerprints, HTTP user agents and open ports to judge whether the device resembles what it claims to be. Forging one MAC is easy; forging an entire device profile consistently is much harder.
Audit the whitelist regularly. Remove MACs for decommissioned equipment promptly. Tie this to your inventory process, and checking vendors on the lookup page will surface whitelist entries that do not match the expected device type.
The effect of address randomisation
Applying MAB to personal mobile devices fails outright once address randomisation is in play, because every connection brings a new address.
The conclusion is direct. Personal devices go through 802.1X with user credentials. MAB is reserved for fixed assets that cannot do 802.1X, and those happen to be exactly the devices that never randomise. That division of labour makes the design both workable and sound.