Use machine learning enabled Anomaly Detection to block zero day threats and other sophisticated attacks. Machine learning automatically and continuously builds and maintains a model of normal user behavior and uses it to identify malicious application traffic. To learn about whether a request is legitimate or a potential malicious attack attempt, it performs the following tasks:
- Captures and collects inputs, such as URL parameters, to build a mathematical model of allowed access
- Matches anomalies against pre-trained threat models
- Detects attacks
Once an anomaly is triggered by the mathematical model, FortiWeb Cloud uses pre-built trained threat models to confirm whether it's a real attack or just a benign anomaly that should be ignored. Each threat model is already trained based on analysis of thousands of attack samples and is continuously updated using the FortiWeb Security Service.
FortiWeb Cloud parses all the URLs in a domain, and builds anomaly detection models for all parameters attached to the URLs.
The anomaly detection model has four stages: collecting, building, testing, and running.
Collecting: To build up an anomaly detection model, the system collects samples of parameter values to train the model.
Building: During the model building stage, the system observes the training samples to self-learn the characteristics of the parameter values and builds up mathematical models.
Testing: After the models are successfully built, they will be tested. All models are required to be tested against a certain number of samples until they have proved to be stable.
Running: When the anomaly detection model is in running state, the system compares parameter value against the anomaly detection model. If it doesn't match the model, the system will record the traffic as an anomaly.
In rare cases, the anomaly detection models for certain parameters are in Discarded stage. It may be caused by several reasons, such as there are too many samples with extremely long parameter value, the system tested the model for three times but all failed, etc.
After anomaly detection model is built, the system will keep on calculating the probability of the new samples and compare it against the model. If the probability of the new samples varies to a large extent for a long period, the system determines this parameter has changed and automatically rebuilds the model based on the new samples.
- Go to SECURITY RULES> Anomaly Detection.
You must have already enabled this module in Add Modules. See How to add or remove a module.
- Configure the following settings.
- Normal Mode: In normal mode, the system collects at least 2500 samples and lasts for at least 7 days. It stops collecting samples after 7 days if at least 2500 samples are collected by then, or continues collecting samples after 7 days until 2500 samples are collected.
- Fast Mode: In fast mode, the sample collecting stops as long as 1500 samples are collected. It's useful when you are testing the model or if there is only small volume of traffic to your application.
- Trust: The system will collect samples only from the IP ranges in the Source IP list.
- Block: The system will collect sample from any IP addresses except the ones in the Source IP list.
- Select the action that FortiWeb Cloud takes when it detects a violation of the rule from the top right corner.
To configure the actions, you must first enable the Advanced Configuration in Global > Settings.
Accept the request and generate a log message.
Alert & Deny
Block the request (or reset the connection) and generate a log message.
- Click SAVE.
|Sample Collection Mode||
|IP List Type||
Whichever option you choose, if you leave the Source IP list blank, the system will collect traffic data samples from any IP address.
Source IP List
Click Create New to list the IP ranges of the samples. Depending on whether you select Trust or Block, FortiWeb Cloud will or will not collect samples from the specified IP ranges.
The Overview tab provides a high level summary of data collected for the domain, including Top 10 URLs by Hit, Violations triggered by anomalies, HMM learning process, Event Dashboard.
The top of the Overview page provides a summary of the data that the machine-learning module has learned about the domain.
Indicates how frequently this application is being accessed.
The date and time when the machine-learning module started to learn about the domain.
The total number of URLs that the machine-learning module has learned.
The total number of block actions that have been triggered since the start time up to the present moment.
The total amount of the HTTP and the HTTPS traffic from the start time up to now.
The charset of URLs in the domain, such as UTF-8.
This chart displays the top 10 URLs for page hits counts.
This chart displays the total number of the potential anomalies and definite anomalies found by the anomaly detection profile.
This chart displays the statistics of HMM learning states of all parameters in the domain. Hover over the circle to check how many parameters are in Collecting, Building, Testing, Running, or Discarded stages respectively. For the explanation of each stage, see Basic concepts.
This chart displays the anomaly detection events, such as sample collection, model running, building and testing, along with the time periods when these events take place.
This tab displays the entire URL directory of the domain in a tree view. You can choose either one of the URLs to view its violation statistics.
The left panel of the Tree View page shows the directory structure of the website. The / (backslash) indicates the root of the site. You can click a URL in the directory tree, then the violation statistics of this URL will be displayed on the right side of the Tree View page. You can also click a directory, then click Relearn Directory or Rebuild Directory to relearn or rebuild anomaly detection models for all the URLs under the selected directory.
This part of the Tree View page shows the statistics of a specific URL.
The frequency at which this URL was accessed in last 24 hours. The frequency is divided into 7 levels, as defined below:
|Model Initialization Date||
The date and time when the mathematical model of this URL was initialized. It shows when FortiWeb began to learn about the data of this URL.
The total number of block actions that have been triggered against this URLsince the start time up to the present moment.
The anomalies detected by the anomaly detection model.
This chart shows the trend of violations in last 24 hours.
The Parameters list shows all the parameters attached to the URL. For example, if the URL is http://www.demo.com/1.php?user_name=jack, then user_name is the parameter. The system builds machine learning model for each parameter, and detects the abnormal parameter values.
You can check the stage of the anomaly detection model of each parameter, and click Rebuild or Relearn if you think the current model does not work.
- Rebuild—Clear the preceding mathematical model for the parameter, and then begin collecting new samples and build the model again. The samples collected for the previous model will be discarded.
- Relearn—Clear the preceding mathematical model for the parameter, and then begin collecting more samples to build the model. The samples collected for the previous model will be not discarded. They will be reused to build the new model.
You can also click Rebuild URL or Relearn URL to rebuild or relearn the models for all the parameters in this URL.