How does the app identify different device models?

GlassShadow
GlassShadow Member Posts: 5
Name Dropper First Comment Photogenic
Hi,
I am working on a project at the moment that helps me with my CEH qualification. Looking into your application as a case study but, cannot for the life of me figure out how it detects the device models.
E.g. I can get that it picks up that the manufacturer might be Google for instance but how does it get that it is a ChromeCast for example.
Thought it might be to do with the MAC addresses or UPNP or NetBIOS but, cant figure out how it gets the information. Didnt know if someone could help me out with this?
Any help would be greatly appreciated!

Best Answer

  • VioletChepil
    VioletChepil London, UKMember Posts: 2,471
    100 Answers 500 Likes 1000 Comments 100 Agrees
    ✭✭✭✭✭✭
    #2 Answer ✓
    Hi @GlassShadow
    Yes definitely you are on the right track here. I believe it is a combination of both of the above. 
    We use machine learning/AI and have a crowd sourced device recognition engine which keeps improving. 

    Community Manager at Fing

Answers

  • VioletChepil
    VioletChepil London, UKMember Posts: 2,471
    100 Answers 500 Likes 1000 Comments 100 Agrees
    ✭✭✭✭✭✭
    Hi @GlassShadow
    Welcome and so interesting you are using Fing App for your case study. 

    The Fing App discovery will identify devices on a best effort basis, based on the information made available by the device, e.g. the MAC address. There are numerous reasons why the device might not be recognized correctly e.g. The DHCP server is not refreshing when speaking to your Fingbox or limited information being offered by specific devices.

    Anything to add @Carlo_from_Fing

    Community Manager at Fing

    GlassShadow
  • Carlo_from_Fing
    Carlo_from_Fing Rome, ItalyAdministrator, Fing Team Posts: 247
    100 Likes 100 Comments 5 Answers 25 Awesomes
    ✭✭✭
    We developed a device recognition technology that uses a combination of classic network protocol fingerprinting (DHCP, MAC, UPnP, Bonjour, etc) with modern AI and Machine Learning.
    This technology is used by the whole Fing product suite and also by our business partners, like e.g. Netgear, as you can read https://www.fing.com/business

    You can also manually try live demo of device recognition at: https://recog.fing.com

    Carlo from Fing

    GlassShadow
  • GlassShadow
    GlassShadow Member Posts: 5
    Name Dropper First Comment Photogenic
    Hi Violet, Hi Carlo,
    Thanks for coming back to me. Okay so I was on the right track, that makes me feel a bit better!
    So I would imagine you have built up some sort of device connection profile database over time using AI and machine learning? Or is it more that different makes/models have specific connection markers/sequences and you had the AI pick up on the differences?
    That's just off the top of my head, if it's much simpler than that or way more complicated that's cool too!
  • rooted
    rooted Gulf Coast, USModerator Posts: 893
    250 Likes 500 Comments 50 Answers Second Anniversary
    ✭✭✭✭✭
    edited December 31, 2019 #6

    It's both, there is a database that uses markers such as MAC address headers to recognize each device. When you enter information into Fing for a device this is crowdsourced to improve recognition.

    Nmap uses similar techniques but isn't as accurate.

    Maya