Introduction to IoT Security
Learn the basics about the underlying mechanisms and fundamental capabilities of the Palo Alto Networks IoT Security solution.
IoT Security is an on-demand cloud subscription service designed to discover and protect the growing number of connected “things” on your network. Unlike IT devices such as laptop computers that perform a wide variety of tasks, IoT devices tend to be purpose-built with a narrowly defined set of functions. As a result, IoT devices generate unique, identifiable patterns of network behavior. Using machine learning and AI, IoT Security recognizes these behaviors and identifies every device on the network, creating a rich, context-aware inventory that’s dynamically maintained and always up to date.
After it identifies a device and establishes a baseline of its normal network activities, it continues monitoring its network activity so it can detect any unusual behavior indicative of an attack or breach. If it detects such behavior, IoT Security notifies administrators through security alerts in the portal and, depending on each administrator’s notification settings, through email and SMS notifications.
IoT Security also uses those behaviors and device identities to automatically generate security policy rule recommendations that allow IoT devices to continue doing normal network activities and block them from doing anything unusual. Panorama or next-generation firewalls can then import these policy rules and enforce them.
For Panorama-managed firewalls that have an IoT Security subscription requiring Cortex Data Lake – Panorama can only import policy rule recommendations if it was used to onboard its managed firewalls to Cortex Data Lake.
The firewall collects metadata from the network traffic of IoT devices, generates Enhanced Application logs (EALs), and forwards them to the logging service. The IoT Security cloud then extracts metadata from these logs for analysis and employs AI and machine-learning algorithms to detect and identify IoT devices using its patented three-tier deep-learning engine:
Tier 1: Device category—IoT Security first identifies the category to which an IoT device belongs. For example, it might identify network behaviors common to all security cameras.
Tier 2: Device profile—IoT Security next constructs a profile of the device, learning its vendor, make, and model. For example, it might discover that the camera behaves in ways that uniquely identify it, such as checking a particular server for software updates for example.
Tier 3: Device instance—IoT Security continues its analysis until it discerns behaviors unique to a specific instance of the identified security camera.
IoT Security looks at over 200 parameters in network traffic metadata, including DHCP option 55 parameter lists, HTTP user agent IDs, protocols, protocol headers, and a host of others. It matches the network traffic patterns of new devices with those of previously identified devices to identify the same types or similar types of devices, even those it is encountering for the first time.
Depending on various factors such as how much network traffic IoT devices generate and how varied their behavior patterns are, IoT Security typically identifies most IoT devices with a high level of confidence during the first day it starts accessing metadata from the logging service. After that, IoT Security continues to increase the number of confidently identified devices until it identifies all or nearly all of them. During this time, you can log in to the IoT Security portal to check that the device inventory is being populated and monitor its progress.
A confidence score indicates the level of confidence IoT Security has in its identification of a device. IoT Security has three confidence levels based on calculated confidence scores: high (90-100%), medium (70-89%), and low (0-69%).
In addition to using machine learning (ML) to observe network traffic and extract various attributes to identify devices and detect anomalous behaviors, IoT Security employs an ML-based model to check for SQL content injected into HTTP URLs, a technique commonly used in SQL vulnerability exploits. By using an ML-based model instead of a model based on rules, IoT Security can find certain patterns of injected SQL content even without specific signatures.
Recommended For You
Recommended videos not found.