search

biometricITC / cPP-CMFA

Repository for work on the CMFA PP-Module and Supporting Documents
0 stars 0 forks source link

Question about "Figure 3. CMFA Score Timeline" #20

Open n-kai opened 2 years ago

n-kai commented 2 years ago

@woodbe

Google's patent (Continuous authentication with a mobile device) *1) includes similar Figure in FIG.3 but this patent doesn't explain math or algorithm how the score can be calculated.

Have you already any specific math or algorithm in your mind that can calculate the score with which the TOE can authenticate user continuously and accurately?

*1) https://patents.google.com/patent/US20150242605A1/en

woodbe commented 2 years ago

The patent is actually for Qualcomm, not Google, so I don't have anything there, but I will poke around to see what I can come up with.

n-kai commented 2 years ago

Thank you. Japanese researcher's main concern is that the CMFA can provide the score anyway but it may be difficult for the CMFA to produce the reliable score so that the mobile can authenticate user continuously and accurately. He think that this may be a reason why there are very few CMFA products in the market.

But he also admitted that his knowledge is limited to the CMFA based on location history gained from the GPS sensor that sometimes produces anomalous data. So the state of art CMFA technology with other types of sensors may produce reliable score.

woodbe commented 2 years ago

What I find interesting in the end is exactly how to do the calculation. I actually expect it would be to individually calculate the components, have them weighted (by relative Trust scores) and then add the result together. So what you would end up with is not an actual "overall" calculation like a biometric, but one where you calculate each individual component based on what is best for that component, and then an additive score as the result.