Multifactor and Multimodal
Most internet users today have hundreds of online accounts, each with a username and password. People simply can’t memorize that many unique, strong passwords so they reuse passwords across accounts, use weak, easy-to-break passwords, or store them where they could be stolen.
Authentication based on something you have may be easier to use but is subject to theft or unauthorized sharing. Biometrics, on the other hand, is both secure (the only factor that verifies the user’s physical presence) and easy to use (nothing to remember).
First of all, user authentication in general simply means making sure you are who you claim to be. We typically speak of three factors of authentication:
- Something you know (password, PIN, mother’s maiden name)
- Something you have (smart card, token, mobile phone)
- Something you are
Biometrics is the “something you are” factor. So biometric authentication means using one or more unique physical (face, voice, fingerprint, etc.) or behavioral traits (voice, keystroke, gait, etc.) to verify who you are.
When greater security is needed biometrics can be combined with other factors to provide “multifactor” or “two-factor” authentication.
Multimodal means that more than one biometric trait is captured, such as face and voice. Strong multimodal biometrics captures all traits at the same time.
Each trait is analyzed separately. The results are combined using a fusion algorithm to provide a single yes/no decision, for instance:
- One of the traits must meet a minimum score
- Every trait must meet a minimum score
- The scores are summed or otherwise mathematically combined and the result must meet a minimum score
This is separate from “multifactor” or “two-factor” or “two-step” authentication, in which one or more biometrics is combined with non-biometric factors such as a password or token.
The benefits of multimodal biometrics are
Accuracy: more accurate than any single biometric trait; high level of convenience AND strong security at the same time.
Fraud resistance: it’s harder to fake multiple traits (especially at the same time) than a single one.
Flexibility: if one trait temporarily changes, for instance your voice changing when you’re sick, the other traits compensate.
User acceptance: higher user acceptance when user is given more than one choice