Automated online policy generation for zero-trust architectures

Inventors: Vitor Lourenco (Niterói, BR); Werner Freund (Rio de Janeiro, BR); Roberto Stelling (Rio de Janeiro, BR); Iam Palatnik (Rio de Janeiro, BR); Vicente Amorim (João Monlevade, BR)

Assignee: DELL PRODUCTS L.P.

G06F21/45

View Patent ↗

Loading inventors, assignments & file history…

Monitor This Case

Get email alerts when status or documents change.

Order Certified Copies

Most orders are placed with the USPTO same day — all within 24 business hours.

Order via The Patent Place →

Pre-filled with this patent's details

Quick Facts

Patent No.: US 12664253
App. No.: 18/310,804
Granted: Jun 23, 2026
Kind: B2

External Links

USPTO Patent Center↗ Full Patent PDF↗

Abstract

Architectures and techniques are described that can automatically (as opposed to manually) generate a new policy in a zero trust architecture (ZTA) or environment, for instance, when it is determined that a current network activity does not have a suitable policy that can be applied, which can be determined by a policy enforcement point (PEP) of a ZTA. To comply with ZTA principles, the new policy can be determined based on data-driven analytics of network activities and can therefore be suitable for online use.

Claims (70)

1 . An online platform employing a zero trust architecture for cyber security, comprising:

at least one processor; and

at least one memory that stores executable instructions that, when executed by the at least one processor, facilitate performance of operations, comprising:

determining, by a policy enforcement point of the zero trust architecture, that new activity data describing a new activity attempting to be performed on the online platform does not have an existing trust policy to apply to the new activity; and

prior to allowing the new activity to be performed on the online platform:

generating, by a graph-based machine learning model, a new embedding of the new activity data, wherein the new embedding conforms to a learned embedding space of an activity knowledge graph comprising known embeddings of known activities, wherein each embedding of the known embeddings is based on a triple comprising a subject of an activity, a predicate of the activity, and an object of the activity, wherein nodes of the activity knowledge graph represent the subjects and the objects, and wherein edges of the activity knowledge graph represent the predicates;

determining linked entity pairs indicative of respective links between the new activity and the known activities;

determining respective probabilities of the linked pairs, wherein the respective probability of the linked pair is based on at least one of:

a first probability of the subject of the new activity and the predicate of the known activity of the linked pair, or

a second probability of the object of the new activity and the predicate of the known activity of the linked pair;

adding a predefined quantity of the linked pairs having highest probabilities to a first order set of top linked pairs;

based on the top linked pairs of the first order set, generating, using a defined process on the activity knowledge graph, a high order set comprising relational path tuples for the known activities of the top linked pairs, wherein each relational path tuple comprises:

the subject of the known activity associated with the relational path tuple,

the object of the known activity associated with the relational path tuple, and

a predicate path comprising a plurality of predicates on a relational path between the subject and the object of the known activity associated with the relational path tuple;

merging the first order set and the high order set to generate a linked entity set;

mapping the linked entity set to candidate trust policies based on known relationships between the activity knowledge graph and clauses of a policy database;

based on one or more clauses of the candidate trust policies, generating a new trust policy for the new activity in compliance with the zero trust architecture; and

enforcing the new trust policy on the new activity attempting to be performed on the online platform.

2 . The online platform of claim 1 , wherein the new activity data is formatted as the triple of the new activity.

3 . The online platform of claim 1 , wherein the learned embedding space is a result of training of the graph-based machine learning model on the activity knowledge graph.

4 . The online platform of claim 3 , wherein the operations further comprise adding the new trust policy to the policy database.

5 . The online platform of claim 1 , wherein the respective probability of the linked pair is further based on a function of a distance between the new activity and the known activity of the linked pair.

6 . The device online platform of claim 1 , wherein the operations further enforcing the new trust policy comprises temporarily employing the new trust policy while the new trust policy is being verified.

7 . The online platform of claim 6 , wherein temporarily employing the new trust policy comprises selectively applying the new trust policy based on a sensitivity of the new activity.

8 . A non-transitory computer-readable medium comprising instructions that, in response to execution, cause an online platform employing a zero trust architecture for cyber security and comprising at least one processor to perform operations, comprising:

prior to allowing the new activity to be performed on the online platform:

generating, by a graph-based machine learning model, a new embedding of the new activity data, wherein the new embedding conforms to a learned embedding space of an activity knowledge graph comprising known embeddings of known activities, wherein each embedding is based on a triple comprising a subject of an activity, a predicate of the activity, and an object of the activity, wherein nodes of the activity knowledge graph represent the subjects and the objects, and wherein edges of the activity knowledge graph represent the predicates;

determining a linked entity group indicative of a group of linked entity pairs indicative of respective links between the new activity and the known activities;