What is AI-Powered Device Profiling?
AI-powered device profiling uses machine learning models that go beyond looking at basic OS settings, signatures, and MAC addresses to accurately identify IoT and smart devices as they connect to wireless and wired networks.
AI-powered device profiling explained
While device profiling is not new, the use of AI and machine learning (ML) in networking to improve the fingerprinting of endpoint clients (i.e., IoT, phones, and laptops) is. With so many new devices being sold today, the accuracy gained by leveraging AI is significant for security and planning purposes.
- As a device (also referred to as clients or endpoints) connects to the network, attributes are collected during the authentication process to create a fingerprint.
- Attributes collected often include information such as the MAC address, operating system version, international mobile equipment identity (IMEI) number, and HTTP user agents.
- Modern solutions now use client and infrastructure telemetry, as well as ML algorithms to dig into how MAC addresses are subdivided by manufacturer and client type. The same can be said for dissecting certain identifiers.
Why AI-powered device profiling?
In the networking space, companies use device profiling and fingerprinting to understand the type and quantity of endpoints that are connected to their networks for security and operating purposes.
Basic use cases:
- For operating purposes, an accurate inventory helps determine what is on the network for bandwidth requirements and troubleshooting.
- From a security perspective, AI-powered device profiling helps to create effective segmentation rules, stop unwanted devices from connecting, and enhance an organizations cybersecurity posture.
Benefits of AI-powered device profiling
The rapid rate of new IoT and smart devices being introduced today makes it more difficult for organizations to identify what is connected to their wired and wireless networks. AI-powered device profiling and ML solutions tackle this by providing firms with improved visibility over their networks.
AI-powered device profiling benefits include:
- The ability to leverage the cloud and data lake to easily capture endpoint attributes and to learn about new IoT devices from the install base.
- Profiling accuracy rates of up to 98% and the ability to decrease the amount of unidentified connection to less than 5%.
- The differentiation of IoT versus BYOD or compute-based clients using recurring application / traffic behavior patterns.
- Cost savings as some vendors include AI-powered device profiling within their foundation management solution and licensing.
How do AI-powered device profiling and traditional device profiling differ?
| Profiling Attribute | AI-powered | Traditional |
|---|---|---|
| Cloud Use | Standard practice | Often on-premises and standalone |
| AI / ML | Yes | No |
| Looks at MAC addresses, OS versions, EMEI, etc. | Yes, with the added benefit of using ML to constantly compare devices for improved accuracy | Yes |
| Crowd sourcing of new fingerprints | Yes, with immediate access to new data across install base | Requires vendors to periodically update software by uploading fingerprints from third-party resources |
| Profiling accuracy | Ranges from 95 to 99% in some scenarios | Ranges from 70 to 80% |
| Rate of “Unknown” device types | Under 5% | Ranges from 30 to 35% |
