Method for creating a zero-trust segmented network perimeter for endpoints or identities

By employing principal clustering analysis and Hadig-Wang clustering techniques, the complexity of network policy creation in zero-trust networking was addressed, enabling automated network resource segmentation and policy template generation, thereby improving the efficiency of network security management.

CN114826642BActive Publication Date: 2026-11-10BLACKBERRY LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202210095948.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2021-01-27
Filing Date
2022-01-26
Publication Date
2026-11-10
Estimated Expiration
2042-01-26

AI Technical Summary

Technical Problem

In zero-trust networking systems, system administrators find it difficult to effectively create network policies or perimeters, especially because they cannot accurately identify the network connection addresses used by applications or services, and end-user behavior is complex and varied, making manual steps cumbersome and time-consuming.

Method used

By acquiring information about classified flows in a computer network, performing master cluster analysis, associating unclassified flows with common services, providing definitive results to the network security management system, creating customized user access policies, and using Hadigan-Wang cluster analysis and flow attributes such as bytes, rate, and DNS name for flow classification and clustering.

Benefits of technology

It enables organizations to receive specific policy templates in a short time, automates network resource segmentation and policy creation, reduces the workload and time cost for administrators, and improves the efficiency of network security management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114826642B_ABST
    Figure CN114826642B_ABST
Patent Text Reader

Abstract

The present disclosure relates to methods for creating a zero-trust segmented network perimeter for endpoints or identities. A method and apparatus for determining services associated with unclassified traffic flows in a computer network is provided. Classification information is obtained for a plurality of classified traffic flows in the computer network, the classification information indicating an association between each of the classified flows and a service. A principal clustering analysis is performed on the plurality of classified flows and the unclassified flow to associate the unclassified flow to a group of classified flows having a common service. The association between the unclassified flow and the common service is provided to a network security management system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application generally relates to network security, and more specifically to the creation of zero-trust segmented network perimeters for endpoints or identities. Background Technology

[0002] In zero-trust networking systems, it can be challenging for enterprise system administrators to create network policies or perimeters for end users. For example, if an administrator wants to allow access to an application or service by an end-user device, the administrator may first need to identify all destination network connections (i.e., Internet Protocol (IP) addresses) used by that application or service.

[0003] This can be problematic because system administrators cannot know all the destination IP addresses used by a given application or service. For example, while an administrator can easily identify certain destination network connections associated with an application or service, the application or service may use other IP addresses unknown to the administrator, such as content delivery networks (CDNs) and / or other third-party services. Similarly, the network destinations associated with certain destination network connections may change periodically, and tracking these destinations can be a lengthy and time-consuming process for system administrators and / or the entire information technology (IT) department.

[0004] Therefore, creating network policies or perimeters for endpoints, identities, and / or end users may require manual steps and may necessitate knowledge of all network services and their associated IP addresses and / or ports. Furthermore, developing effective network policies for endpoints, identities, and / or end users requires administrators to understand what type of work or organizational part a user might be involved in in order to understand what network services the user might need to access. Attached Figure Description

[0005] Reference will now be made to the accompanying drawings, which illustrate exemplary embodiments of this application, by way of example, wherein:

[0006] Figure 1 A simplified example of a computing device is shown in block diagram form;

[0007] Figure 2 Example systems in which the methods and apparatus according to this specification can be implemented are shown;

[0008] Figure 3 An example method for identifying services associated with unclassified flows is shown in the form of a flowchart;

[0009] Figure 4 An example of data processing and unsupervised classification according to one embodiment is shown; and

[0010] Figure 5An example administrator graphical user interface (GUI) is shown, illustrating the association between clustering flows and services.

[0011] Similar reference numerals are used in different accompanying drawings to indicate similar components. Detailed Implementation

[0012] abbreviations

[0013] Various abbreviations are used in this disclosure. The following are definitions of the most commonly used abbreviations that may appear throughout this disclosure:

[0014] • AD: Active Directory

[0015] CDN: Content Delivery Network

[0016] DNS: Domain Name System

[0017] GUI: Graphical User Interface

[0018] • IaaS: Infrastructure as a Service

[0019] IP: Internet Protocol

[0020] IT: Information Technology

[0021] MiTM: Middleman

[0022] SaaS: Software as a Service

[0023] • SNI: Server Name Indicator

[0024] •TLS: Transport Layer Security

[0025] In a first aspect, this application describes a method for determining a computer implementation of a service associated with an unclassified traffic flow in a computer network. The method may include: obtaining classification information for a plurality of classified traffic flows in the computer network, wherein the classification information indicates an association between each of the plurality of classified flows and one of a plurality of services. The method may further include: performing a master clustering analysis on the plurality of classified flows and the unclassified flows to associate the unclassified flows with a group of classified flows having a common service; determining that the unclassified flows are associated with the common service; and providing the determination to a network security management system.

[0026] In some implementations, the network security management system combines this determination with user network access grouping information to create a baseline network security profile for one or more user access groups. In other implementations, the network security management system creates a customized user profile based on the intersection or union of the baseline network security profiles associated with the user.

[0027] In some implementations, providing this determination to a network security management system includes modifying user access policies to whitelist Internet Protocol (IP) addresses associated with unclassified flows.

[0028] In some implementations, classified traffic flows can be encrypted. In some implementations, unclassified traffic flows can be Transport Layer Security (TLS) flows.

[0029] In some implementations, performing primary clustering analysis involves partitioning flows based on one or more flow attributes. In some implementations, flow attributes may include one or more of the following: bytes transmitted, packets transmitted, data rate transmitted, packet rate transmitted, packet size per packet, duration of bidirectional flow, bytes received, packets received, downlink data rate, downlink packet rate, packet size per packet received, time parameter "t" until idle, and Domain Name System (DNS) name. The time parameter "t" is a time window within which subsequent network flows have occurred.

[0030] In some implementations, the clustering analysis is k-means clustering, such as Hartigan-Wong clustering. In other implementations, the main clustering analysis is based on preliminary clustering analysis applied to multiple classified flows and the corresponding users associated with each classified flow within those flows.

[0031] In another aspect, this application describes a non-transitory computer-readable storage medium storing processor-readable instructions that, when executed, configure the processor to perform any of the methods described herein. This application also describes a computing device comprising: a processor, memory, and an application containing processor-executable instructions that, when executed, cause the processor to perform at least one of the methods described herein. In this regard, the term processor is intended to include all types of processing circuitry or chips capable of executing program instructions.

[0032] Other aspects and features of this application will be understood by those skilled in the art by reading the following illustrative description in conjunction with the accompanying drawings.

[0033] In this application, the terms “about,” “approximately,” and “substantially” are intended to cover variations (such as changes in properties, parameters, and dimensions) that may exist within the upper and lower limits of the numerical range. In non-limiting instances, the terms “about,” “approximately,” and “substantially” may mean plus or minus 10% or less.

[0034] In this application, the term "and / or" is intended to cover all possible combinations and sub-combinations of the listed elements, including any single element, any sub-combination, or all elements listed, and does not necessarily exclude additional elements.

[0035] In this application, the phrase “...or at least one of ...” is intended to cover any one or more of the listed elements, include any one of the listed elements, any sub-combination or all elements, without necessarily excluding any additional elements, and without necessarily requiring all elements.

[0036] While many of the example computing devices described and discussed below are mobile devices, such as smartphones, tablets, and laptops, this application is not limited to mobile devices but can be implemented in combination with non-mobile devices.

[0037] As applications continue to move to the cloud, security is becoming an increasing concern, where endpoint devices can access cloud-based applications from anywhere in the world. In fact, applications hosted in the cloud are particularly vulnerable to malicious attacks on protected data.

[0038] Zero-trust networking is a relatively new paradigm seeking to address some of the security challenges associated with cloud-based applications. Cloud-based applications can include software as a service (SaaS), enterprise applications utilizing Infrastructure as a Service (IaaS), and enterprise applications hosted in data centers. Typically, zero-trust networking operates under the principle of defaulting to untrust across all networks, including both private enterprise networks and external public networks. Therefore, for example, even if a malicious entity wants to access endpoint devices in a private enterprise network, lateral movement within the network (e.g., from an infected device to a server) can be prevented due to the segmentation of network resources by applying zero-trust networking principles. However, while effective in many cases, zero-trust networking also employs relatively restricted methods to allow access to network resources. Because many cloud applications periodically move content from one web server to another, endpoint devices (end users) may discover that they are blocked from accessing content necessary for the application until the network identifies the new web server as trusted.

[0039] A system and methodology that learns end-user network access behavior and can group flows into application classes or services (thus allowing administrators to create network perimeters or network policies) is highly desirable. Similarly, systems and methods capable of classifying network traffic, understanding end-user context (e.g., which Active Directory a user belongs to), and automatically segmenting network profiles (which can be applied as policies) are also highly desirable. Furthermore, achieving these results without requiring the decryption of network traffic is also highly desirable.

[0040] Now for reference Figure 1 , Figure 1 A simplified example of computing device 100 is shown in block diagram form. Computing device 100 includes a processor 102, memory 104, a user interface 106, and one or more input devices 108. Memory 104 may include temporary memory and permanent memory. In many embodiments, user interface 106 may include a display screen, but may also include other sensing output devices, including speakers, vibration mechanisms, electronic ink displays, or other such devices. Input devices 108 may include touch-sensitive input devices such as touchscreens, keyboards, keypads, cameras, microphones, gyroscopes, accelerometers, and various other input mechanisms.

[0041] The computing device 100 may also include various sensors 110 and one or more communication subsystems 112, some of which may be input devices 108. The communication subsystem 112 enables wired or wireless communication with other computing devices. An exemplary communication subsystem 112 includes a cellular system for data and / or voice communication over a cellular network (including 3G, 4G, or 5G networks), a Wi-Fi chip for data connectivity to a wireless local area network (WLAN), and Bluetooth. TM The chip is a short-range wireless system, a near-field communication (NFC) chip, an Ethernet port, or any other communication system operating according to any applicable communication protocol.

[0042] Now for reference Figure 2 The illustration shows an example system 200 in which the methods and apparatus according to this specification can be implemented. A trusted endpoint 202 is configured to communicate with other network destinations 206 (e.g., a web server) via a computer network 204. The computer network 204 may include a wired network, a wireless network, or a combination thereof, and may include the Internet. In some embodiments, the trusted endpoint 202 may be a mobile device, such as a smartphone, smartwatch, laptop computer, tablet computer, or wearable device, or may be a non-mobile device, such as a desktop computing device.

[0043] Network 204 can be connected to flow analysis application 210. In some embodiments, flow analysis application 210 may reside on server 208. Flow analysis application 210 may be configured to communicate with network management system 212. In some embodiments, network management system 212 and flow analysis application 210 may reside on the same server. In some embodiments, flow analysis application 210 and network security management system 212 may reside on different servers. In some embodiments, network security management system 212 may include multiple servers.

[0044] Many organizations exercise enterprise control over computing devices and the resources accessible by those devices. Organizations may exercise specific control over particular applications or sets of applications. Organizations may optionally or additionally manage entire endpoint devices or profiles on endpoint devices. Generally, an organization can manage the ability of computing devices to access the corporate network and can push device policies and security policies to these devices. As a result of this control, the enterprise can monitor network traffic at endpoint devices and determine which endpoint devices are "trusted."

[0045] In a zero-trust network, all network resources can be considered untrusted by default. As a result, one or more trusted endpoints 202 can attempt to connect to one or more network destinations 206. The trusted endpoints 202 can make these attempts via originating applications or services permitted by the network security management system, such as, for example, Office 365. TM and / or Salesforce TM Alternatively, the trusted endpoint 202 may make these attempts to connect directly or via an application (such as a gambling application) that is not permitted by the network management system. Attempts to connect from an endpoint to an untrusted network destination may result in one or more bidirectional connections or flows 216. The flow analysis application 210 may then attempt to classify these flows 216 to the originating application of these flows. Classification techniques may include, for example, using a 5-tuple (source IP address, source port, destination IP address, destination port, and transport protocol) of network packets flowing in either direction. Classification techniques may also include examining the unencrypted portions of the TLS connection handshake during connection establishment, such as the exchanged certificates, Server Name Indication (SNI) data, and Domain Name System (DNS).

[0046] The classification performed by the flow analysis application 210 can result in the identification of one or more classified flows and one or more unclassified flows. To associate unclassified flows with their originating applications or services, clustering algorithms can be applied to the flows. Clustering algorithms can be applied in conjunction with the temporal order within the flow's time series.

[0047] Flow clustering may require correctly grouping flows for each application class into groups based on various flow attributes. These flow attributes may include the number of bytes transmitted, the number of packets transmitted, the data rate transmitted, the packet rate transmitted, the packet size transmitted per packet, the duration of bidirectional flows, the number of bytes received, the number of packets received, the downlink data rate, the downlink packet rate, the packet size received per packet, the idle time parameter "t", and the DNS name. Therefore, a vector identifying the cluster can be represented per application F based on the above parameters.

[0048] Cluster analysis of the flows can then be applied. In some embodiments, a computationally efficient Hadigan-Wang implementation of cluster analysis can be implemented for k-means flow cluster analysis. In some embodiments, another k-means cluster analysis can be implemented. Once segmented and clustered, these flows can then be automatically tagged with their corresponding enterprise application classes or services; for example, Microsoft. TM Services, Salesforce TM Service or Workday TM Service. Tagging can occur automatically based on the deviation of each flow from the relevant SNI / DNS for the cluster name.

[0049] The named TLS flow clusters can then be provided to the network security management system 212. The network security management system 212 can then enrich the flow clusters using Active Directory (AD) grouping. The network security management system 212 can modify one or more user access policies to whitelist one or more IP addresses associated with one or more named flow clusters. As a result, customized policy templates can be created for each user or user group within the organization. Administrators can then select these policies and / or create exclusions based on them. Using named AD groups to profile clustered flows helps administrators visualize the types of applications used within the enterprise (including shadow IT applications).

[0050] The advantage of the above method is that it can provide enterprise-specific strategy templates after being executed by a small number of users for only a short period of time.

[0051] Now for reference Figure 3 , Figure 3 An example method 300 for identifying services associated with unclassified traffic flows in a computer network is illustrated in flowchart form. Method 300 can be implemented by processor-executable instructions that, when executed by one or more processors, cause the processors to perform the described operations. Example operations or functions are described below.

[0052] In operation 302, classification information is obtained for multiple classified flows. The classification information may include the association between each of the multiple classified flows and one of multiple applications or services. For example, multiple flows may be observed between a user device or endpoint and multiple destination IP addresses. This is to determine which of these flows are associated with an application or service that allows the user access (e.g., Office 365). TM or Salesforce TM These streams can be categorized by applications or services.

[0053] Historically, stream classification has been achieved through applications that use business decryption. However, with the increasing use of end-to-end encryption services in applications (such as TLS authentication and the adoption of TLS 1.3), business decryption, including methods based on MiTM, may no longer be possible. Similarly, privacy is becoming increasingly important, and inspecting content as it flows through the network can have legal implications.

[0054] The first step in creating a policy is to group connection sets and identify the application or service associated with each connection set. A common approach for classifying flows is by evaluating the associated 5-tuple information (source IP address, destination IP address, source port, destination port, and protocol). In some embodiments, where the flow is a TLS flow, further classification can be obtained by examining the unencrypted portion of the TLS connection handshake, including looking for exchanged certificates, SNI data, and DNS. Once the destination is classified, various business flow policies can be applied to it. However, it may not be easy to detect applications associated with a business based solely on network 5-tuple information and / or observing the unencrypted portion of the connection. A single application on a mobile or desktop device may require many connections and many flows; there may be N flows that define a single application. Because this type of classification can successfully classify some but not all flows for a given application, additional analysis may be required.

[0055] In operation 304, a master clustering analysis is performed on multiple classified and unclassified flows. Flow clustering involves correctly grouping flows for each application class into groups based on various flow attributes. Examples of flow attributes that can be used for this purpose include: bytes transmitted, packets transmitted, data rate transmitted, packet rate transmitted, packet size transmitted per packet, duration of bidirectional flow, bytes received, packets received, downlink data rate, downlink packet rate, packet size received per packet, time to idle parameter "t", and DNS name. Therefore, based on the above parameters, a vector identifying the cluster can be represented for each application F.

[0056] Primary clustering analysis can include k-means clustering analysis, such as the computationally efficient "Hadigen and King" implementation of clustering analysis. Separation of clusters can be examined. For example, clustering can be performed on both network services and users of those services. Clustering can be implemented in two steps: corresponding to N-dimensional clustering in a 2-relation space, there can be initial clustering of services, followed by primary clustering of end-user usage of those services. In other words, there can be initial clustering of services, followed by primary clustering of end-users using those services. End-user-based flow clustering can provide associations between applications or services and one or more AD groups. Referring to step 306, once segmented and clustered, these flows can be automatically tagged with the associated enterprise application classes or services. Examples of enterprise application classes or services include, for example, Microsoft... TM Services, Salesforce TM Service and Workday TM Service. Tagging can occur automatically based on the deviation of each flow from the relevant SNI / DNS for the cluster name.

[0057] At step 308, the identification of services associated with the unclassified flows is provided to the network security management system. These named flow clusters can then be enriched using user AD grouping information. This allows for the creation of customized policy templates for each user or user group within the organization. This enables administrators to select policies or create exclusions based on the same policies. Analyzing clustered flows using named AD groups can also help administrators visualize the types of applications used within the enterprise (including shadow IT applications).

[0058] Figure 4 Data processing, unsupervised classification, and cluster analysis according to one embodiment are illustrated. At 402, a network traffic collector receives and analyzes network traffic information. This network traffic information can be received from sources such as network flows and / or pcaps. At step 406, information associating users and / or endpoints with one or more connections or flows can be provided. Classification of connections or flows can be performed at step 408. At step 410, information associating flows with corresponding applications 410 is then provided at step 412 for cluster analysis. Information 410 may also include indications of one or more flows without a corresponding association (i.e., unclassified flows). As a result of cluster analysis 412, one or more unclassified flows can be associated with an application or service. The result of cluster analysis 412 can then be provided to a network security management system, where it can be used to provide enterprise network policy templates for one or more end users 414.

[0059] Figure 5A graphical user interface (GUI) 500, according to one embodiment, is illustrated and can be used to provide flow classification results to one or more administrators. Column 502 provides a list of multiple different network flows. Column 504 provides a list of applications or services associated with the various network flows, which can be identified through flow classification and flow clustering. According to some embodiments, administrators can use GUI 500 to select network security policies or create exclusions based on network security policies.

[0060] The exemplary embodiments of this application are not limited to any particular operating system, system architecture, mobile device architecture, server architecture, or computer programming language.

[0061] It should be understood that applications, modules, routines, processes, threads, or other software components implementing the described methods / processes can be implemented using standard computer programming techniques and languages. This application is not limited to specific processors, computer languages, computer programming conventions, data structures, or other such implementation details. Those skilled in the art will recognize that the described processes can be implemented as part of computer-executable code stored in volatile or non-volatile memory, as part of an application-specific integrated circuit (ASIC), etc.

[0062] Certain adjustments and modifications can be made to the described embodiments. Therefore, the above embodiments are considered illustrative rather than restrictive.

Claims

1. A method for determining a computer implementation of a service associated with an unclassified traffic flow in a computer network, the method comprising: Obtain classification information for multiple classified service flows in the computer network, wherein the classification information indicates the association between each of the multiple classified service flows and one of a plurality of services; Perform a master clustering analysis on the plurality of classified service flows, the corresponding users associated with each of the plurality of classified service flows, and the unclassified service flows to associate the unclassified service flows with a group of classified service flows that have a common service; Determine that the unclassified service flow is associated with the common service; as well as The determination is provided to the network security management system for combination with user network access grouping information to create a network security profile for one or more user access groups.

2. The computer-implemented method of claim 1, wherein the network security management system creates a customized user profile based on the intersection or union of the network security profiles associated with the user.

3. The computer-implemented method according to claim 1, wherein creating the network security profile includes: Modify user access policies to whitelist Internet Protocol (IP) addresses associated with the unclassified service flows.

4. The computer-implemented method according to claim 1, wherein the unclassified traffic flow is encrypted.

5. The computer-implemented method according to claim 1, wherein the unclassified traffic flow and the plurality of classified traffic flows are encrypted.

6. The computer-implemented method of claim 1, wherein the main clustering analysis is based on preliminary clustering analysis applied to the plurality of classified traffic flows and the corresponding user associated with each of the plurality of classified traffic flows.

7. The computer-implemented method according to claim 1, wherein the principal clustering analysis is k-means clustering analysis.

8. The computer-implemented method according to claim 7, wherein the principal clustering analysis is Hadigen-Wang clustering analysis.

9. The computer-implemented method of claim 1, wherein the unclassified traffic flow is a Transport Layer Security (TLS) flow.

10. The computer-implemented method of claim 1, wherein performing cluster analysis comprises: Streams are divided based on one or more stream attributes.

11. The computer-implemented method of claim 10, wherein the one or more stream attributes include at least one attribute selected from the group consisting of: bytes transmitted, packets transmitted, data rate transmitted, packet rate transmitted, packet size transmitted per packet, duration of bidirectional stream, bytes received, packets received, downlink data rate, downlink packet rate, packet size received per packet, and time to idle parameter. t And the Domain Name System (DNS) name.

12. A computing device, comprising: processor; Memory; The analysis application, stored in the memory, contains processor-executable instructions that, when executed by the processor, cause the processor to determine services associated with unclassified traffic flows in a computer network, wherein the processor-executable instructions cause the processor to: Obtain classification information for multiple classified service flows in the computer network; wherein the classification information indicates the association between each of the multiple classified service flows and one of a plurality of services; Perform a master clustering analysis on the plurality of classified service flows, the corresponding users associated with each of the plurality of classified service flows, and the unclassified service flows to associate the unclassified service flows with a group of classified service flows that have a common service; Determine that the unclassified service flow is associated with the common service; as well as The determination is provided to the network security management system for combination with user network access grouping information to create a network security profile for one or more user access groups.

13. The computing device of claim 12, wherein creating the network security profile comprises: Modify user access policies to whitelist Internet Protocol (IP) addresses associated with the unclassified service flows.

14. The computing device of claim 12, wherein the unclassified traffic flow is encrypted.

15. The computing device of claim 12, wherein the unclassified traffic flow and the plurality of classified traffic flows are encrypted.

16. The computing device of claim 12, wherein performing cluster analysis includes partitioning the streams based on one or more stream attributes.

17. The computing device of claim 16, wherein the one or more stream attributes include at least one attribute selected from the group consisting of: bytes transmitted, packets transmitted, data rate transmitted, packet rate transmitted, packet size transmitted per packet, duration of bidirectional stream, bytes received, packets received, downlink data rate, downlink packet rate, packet size received per packet, and time to idle parameter. t And the Domain Name System (DNS) name.

18. The computing device of claim 12, wherein the main clustering analysis is based on a preliminary clustering analysis applied to the plurality of classified traffic flows and the corresponding user associated with each of the plurality of classified traffic flows.

19. A non-transitory computer-readable storage medium storing processor-executable instructions for determining a service associated with an unclassified traffic flow in a computer network, wherein the processor-executable instructions, when executed by a processor, cause the processor to: Obtain classification information for multiple classified service flows in the computer network; wherein the classification information indicates the association between each of the multiple classified service flows and one of a plurality of services; Perform a master clustering analysis on the plurality of classified service flows, the corresponding users associated with each of the plurality of classified service flows, and the unclassified service flows to associate the unclassified service flows with a group of classified service flows that have a common service; Determine that the unclassified service flow is associated with the common service; as well as The determination is provided to the network security management system for combination with user network access grouping information to create a network security profile for one or more user access groups.

Citation Information

Patent Citations

  • Internet flow distinguishing method

    CN102685016A

  • Classification of unknown network traffic

    US20200366717A1