# US 2019/0377872 A1 · Multiplexed - proactive resiliency system

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2019/0377872 A1
- Document ID: US20190377872A1
- Status: Published application
- Kind: A1
- Application number: 16/545,179
- Filed: 2019-08-20
- Published: 2019-12-12
- Earliest claimed: 2017-09-11
- Assignees: Bank of America Corporation
- Applicants: Bank of America Corporation
- Inventors: Sasidhar Purushothaman
- Web page: https://patentgrep.com/patent/US20190377872A1

## CPC

- G06F 21/554: {involving event detection and direct action}
- G06F 21/552: {involving long-term monitoring or reporting}
- H04L 63/1425: {Traffic logging, e.g. anomaly detection}
- H04L 63/145: {the attack involving the propagation of malware through the network, e.g. viruses, trojans or worms}
- G06F 2221/033: Test or assess software
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

Methods for machine-learned detection and removal of malicious software within a network are provided. Methods may record environment behavior of an application and a plurality of components. The plurality of components may touch the application. Methods may generate a baseline dataset based on the recorded environment behavior. Methods may schedule snapshots of the application. Methods may take snapshots of the application and the components based on the scheduling. Methods may store the snapshots in a repository. Methods may monitor the application and the components, using the stored snapshots, for any deviation in the environment behavior. Methods may detect a deviation in the behavior of the application or components. Methods may take a snapshot, outside of the scheduling, of the application and components upon detection of the deviation. Methods may determine that the deviation is unwarranted. Methods may revert the application and components back to a previous version.

## Claims

### Claim 1

A method for machine-learned detection and removal of malicious software within a network, the method comprising:
receiving a request for tracing a network application:
allocating memory in a trace data repository for tracing the network application:
recording, within the trace data repository, environment behavior of:
the network application; and
a plurality of network components that touch communicate with the application, said network components comprising software layers, software dependencies, libraries and hardware dependencies:
identifying each of the software layers, software dependencies, libraries and hardware dependencies using a unique identifier;
combining the unique identifiers into an application reference identifier, said application reference identifier comprising the combination of network components that communicates with the application and the order in which the application communicates with the network components;
based on the recorded environment behavior, generating a baseline dataset within the trace data repository,
scheduling a plurality of snapshots of the application, each of the snapshots occurring at a predetermined periodic interval;
capturing a first snapshot of the application and the plurality of components, the first snapshot corresponding to one of the plurality of scheduled snapshots;
storing the first snapshot in the trace data repository;
monitoring the application and the plurality of components, using the stored snapshots and the baseline dataset, for any deviation in the environment behavior;
detecting a first deviation in the environment behavior of the application or the plurality of components;
in response to detecting a first deviation, capturing a second snapshot of the application and the plurality of components, the second snapshot being inconsistent with the plurality of scheduled snapshots;
transmitting the second snapshot as an alert to one or more stakeholders associated with the application;
receiving a flag from the one or more stakeholders relating to the first deviation, the flag identifying the first deviation as intended or unwarranted;
detecting a second deviation included in the application or the plurality of components;
identifying a second deviation as intended or unwarranted based on the received flag;
upon determining that the second deviation is unwarranted, and, therefore, is directed to malicious software, reverting, using the application reference identifier, the application and the plurality of components back to a previous version of the application and the plurality of components, thereby removing the malicious software; and
upon determining that the second deviation is intended, storing the intended deviation in a log of verified intended deviations.

### Claim 2 (depends on 1)

The method of claim 1, wherein the scheduling is based on a level of criticality associated with the application.

### Claim 3 (depends on 1)

The method of claim 1, wherein the previous version of the application and the plurality of components is the most recent snapshot, prior to the second deviation, of the application and the plurality of components.

### Claim 4 (depends on 1)

The method of claim 1, wherein the stored snapshots are used to identify issues and their causes, associated with the application and the plurality of components.

### Claim 5 (depends on 1)

The method of claim 1, wherein the stored snapshots are used to enhance the performance of a transmitted new application, such that an engine learns from the recorded environment behavior to simulate at least one new application and determine a suitable set of tools, components, code routines and/or environment for hosting the new application.

### Claim 6 (depends on 1)

The method of claim 1, wherein the stored snapshots are used to identify and determine a composite security breach exposure metric of an environment, said environment that includes the application and the plurality of components, said composite security breach exposure metric corresponding to the sum of a plurality of security breach exposure metrics, each security breach exposure metric corresponding to one of the plurality components.

### Claim 7 (depends on 1)

The method of claim 1, wherein the stored snapshots are used to appropriately allocate resources within the network.

### Claim 8

An apparatus for machine-learned detection and removal of malicious software within a network, the apparatus comprising:
a trace data repository, the trace data repository configured to:
receive a request for tracing a network application;
allocate memory for tracing the network application;
record environment behavior of:
the network application; and
a plurality of network components that communicate with the application, said network components comprising software layers, software dependencies, libraries and hardware dependencies;
identify each of the software layers, software dependencies, libraries and hardware dependencies using a unique identifier;
combine the unique identifiers into an application reference identifier, said application reference identifier comprising the combination of network components that communicates with the application and the order in which the application communicates with the network components;
a processor configured to:
generate a baseline dataset based on the recorded environment behavior; and
store the baseline dataset in the trace data repository;
a hardware-processor-scheduler configured to generate a schedule for capturing a plurality of substantially simultaneous snapshots of the application and the plurality of components at a plurality of predetermined periodic intervals;
the processor further configured to:
capture a plurality of simultaneous snapshots of the application and the plurality of components according to the schedule;
store the plurality of captured snapshots in the trace data repository;
monitor, using the stored snapshots and the baseline dataset, the application and the plurality of components, for any deviation in the environment behavior of the application or the plurality of components;
detect a deviation in the environment behavior of the application or in at least one of the plurality of components;
in response to detecting the deviation, capture a second simultaneous snapshot of the application and the plurality of components, the second simultaneous snapshot being inconsistent with the schedule;
determine, based on previously recorded snapshots, whether the deviation is intended or unwarranted;
upon determination that the deviation is unwarranted and, therefore, is caused by malicious software, revert, using the application reference identifier, the application and the plurality of components to a previous version of the application and the plurality of components, thereby removing the malicious software; and
upon determination that the deviation is intended, storing the intended deviation in a log of verified intended deviations.

### Claim 9 (depends on 8)

The apparatus of claim 8, wherein the scheduler generates the schedule based on a level of criticality associated with the application.

### Claim 10 (depends on 8)

The apparatus of claim 8, wherein the previous version of the application and the plurality of components is the most recent snapshot, prior to the deviation, of the application and the plurality of components.

### Claim 11 (depends on 8)

The apparatus of claim 8, wherein the stored snapshots are used to identify issues and their causes, associated with the application and the plurality of components.

### Claim 12 (depends on 8)

The apparatus of claim 8, wherein the stored snapshots are used to enhance the performance of a transmitted new application, such that an engine learns from the recorded environment behavior to simulate at least one new application and determine a suitable set of tools, components, code routines and/or environment for hosting the new application.

### Claim 13 (depends on 8)

The apparatus of claim 8, wherein the stored snapshots are used to identify and determine a composite security breach exposure metric of an environment, said environment that includes the application and the plurality of components, said composite security breach exposure metric corresponding to the sum of a plurality of composite security breach exposure metrics, each security breach exposure metric corresponding to one of the plurality of components.

### Claim 14 (depends on 8)

The apparatus of claim 8, wherein the stored snapshots are used to appropriately allocate resources within the network.

### Claim 15

A method for machine-learned detection and removal of malicious software within a network, the method comprising:
receiving a request for tracking a network application;
allocating memory in a trace data repository for tracing the network application; recording, within a trace data repository, environment behavior of:
the network application; and
a plurality of network components that communicate with the application, said network components comprising software layers, software dependencies, libraries and hardware dependencies;
identifying each of software layers, software dependencies, libraries and hardware dependencies using a unique identifier;
combining the unique identifiers into an application reference identifier, said application reference identifier comprising the combination of network components that communicates with the application and the order in which the application communicates with the network components;
based on the recorded environment behavior, generating a baseline dataset within the trace data repository;
scheduling a plurality of snapshots of the application, each of the snapshots occurring at a predetermined periodic interval;
capturing a first snapshot of the application and the plurality of components, the first snapshot corresponding to one of the plurality of scheduled snapshots:
storing the first snapshot and data associated with the first snapshot in the trace data repository, said data comprising:
a trace identification sequence identifying the application;
an infra reference identifier identifying an environment setup of the application at the time of the first snapshot;
an application span reference identifier identifying the plurality of components that communicate with the application at the time of the first snapshot;
a code reference identifying a static reference to a deployed piece of code, said deployed piece of code being the basis for functioning of the application;
performing a simulated restoration of the application and the plurality of components back to the first snapshot;
based on the simulated restoration, determining a confidence level tor recovering the application and the plurality of components;
storing, in a restoration reference repository:
an iteration identifier identifying the first snapshot;
a recoverability metric identifying the confidence level;
a validated status identifying whether the recoverability metric has been validated;
monitoring the application and the plurality of components, using the trace data repository, for any deviation in the environment behavior;
detecting a deviation in the environment behavior of the application of the plurality of components;
in response to detecting the deviation, receiving a flag relating to the deviation;
based on the received flag, identifying the deviation as intended or unwarranted;
determining that the deviation is unwarranted and, therefore, is directed to malicious software;
upon determining that the deviation is unwarranted, reverting, using the application reference identifier, the application and the plurality of components back to a previous version of the application and the plurality of components, thereby removing the malicious software, said previous version being the most recent snapshot in which the recoverability metric is above a predetermined figure; and
upon determining that the deviation is intended, storing the intended deviation in a log of verified intended deviations.

### Claim 16 (depends on 15)

The method of claim 15, wherein the scheduling is based on a level of criticality associated with the application.

### Claim 17 (depends on 15)

The method of claim 15, wherein the stored snapshots are used to identify issues and their causes, associated with the application and the plurality of components.

### Claim 18 (depends on 15)

The method of claim 15, wherein the stored snapshots are used to enhance the performance of a transmitted new application, such that an engine learns from the recorded environment behavior to simulate at least one new application and determine a suitable set of tools, components, code routines and/or environment for hosting the new application.

### Claim 19 (depends on 15)

The method of claim 15, wherein the stored snapshots are used to identify and determine a composite security breach exposure metric of an environment, said environment that includes the application and the plurality of components, said composite security breach exposure metric corresponding to the sum of a plurality of security breach exposure metrics, each security breach exposure metric corresponding to one of the plurality components.

### Claim 20 (depends on 15)

The method of claim 15, wherein the stored snapshots are used to appropriately allocate resources within the network.
