# US 2023/0412623 A1 · Cyberattack detection with topological data

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

- Number: US 2023/0412623 A1
- Document ID: US20230412623A1
- Status: Published application
- Kind: A1
- Application number: 17/844,429
- Filed: 2022-06-20
- Published: 2023-12-21
- Earliest claimed: 2022-06-20
- Assignees: Raytheon Company
- Applicants: Raytheon Company
- Inventors: Nandi O. Leslie
- Web page: https://patentgrep.com/patent/US20230412623A1

## CPC

- H04L 63/1425: {Traffic logging, e.g. anomaly detection}
- H04L 63/1458: {Denial of Service}
- H04L 2463/144: Detection or countermeasures against botnets
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

Discussed herein are devices, systems, and methods for detecting anomalous or malicious processes based on a network flow data. A method for network intrusion detection, the method comprising receiving a network flow data, implementing a topological data analysis (TDA) algorithm to identify respective birth and death of homological classes to which the network flow data maps, appending the respective (birth, death) pairs along with additional TDA-based features to a feature space resulting in an augmented feature space, and determining, using a machine learning algorithm the operates on the augmented feature space as input, whether the network flow data is associated with a network intrusion.

## Claims

### Claim 1

A device of a computer network, the device comprising:
at least one memory including instructions stored thereon; and
processing circuitry configured to execute the instructions, the instructions, when executed, cause the processing circuitry to perform operations comprising:
receiving a network flow data;
implementing a topological algorithm to identify respective (birth, death) pairs to which the network flow data maps;
appending the respective (birth, death) pairs to a feature vector representing the network flow data resulting in an augmented feature vector; and
determining, using a classifier that operates using the augmented feature vector as input, whether the network flow data is associated with a network intrusion.

### Claim 2 (depends on 1)

The device of claim 1, wherein the network intrusion includes a botnet, denial of service (DoS), distributed DoS (DDoS), brute force, structured query language (SQL) injection, infiltration, or port scan.

### Claim 3 (depends on 1)

The device of claim 1, wherein the classifier is a clustering classifier that clusters the augmented feature vectors into groups including a first group corresponding to a benign network flow data and a second group corresponding to a malicious network flow data.

### Claim 4 (depends on 1)

The device of claim 1, wherein the classifier performs K-means clustering or k-nearest neighbors (k-NN) clustering.

### Claim 5 (depends on 1)

The device of claim 1, wherein the classifier is a multilayer perceptron (MLP).

### Claim 6 (depends on 1)

The device of claim 1, wherein data of the network flow data includes three or more of an input interface index, an output interface index, timestamps for flow start and finish times, number of bytes and packets observed in a corresponding flow, source and destination internet protocol (IP) address, internet control message protocol (ICMP) type and code, IP protocol, type of service value, source and destination port numbers for transmission control protocol (TCP), unigram data protocol (UDP), or stream control transmission protocol (SCTP), union of all TCP flags observed over a life of the flow, IP address of an immediate next-hop along a route to a destination, or source and destination IP masks.

### Claim 7 (depends on 6)

The device of claim 6, wherein the operations further comprise:
extracting features of the network flow data; and
wherein the topological algorithm operates on the extracted features to generate a persistence diagram for the network flow data along with other network flow data.

### Claim 8 (depends on 7)

The device of claim 7, further comprising, before extracting the features, completing the network flow data to alter text to numerical values and fill-in any empty fields of the network flow data.

### Claim 9 (depends on 8)

The device of claim 8, further comprising, after completing the network flow data, balancing a number of samples of each class in network flow data including the network flow data.

### Claim 10

A non-transitory machine-readable medium including instructions that, when executed by a machine, cause the machine to perform operations for network intrusion detection, the operations comprising:
receiving a network flow data;
implementing a TDA algorithm for the persistence diagram to identify respective (birth, death) diameter pairs to which the network flow data maps;
appending the respective (birth, death) pairs and other topological features, including an identifier for each d-dimensional simplex formed for a given parameter to a feature vector representing the network flow data resulting in an augmented feature vector; and
determining, using a classifier or clustering algorithm that operates using the augmented feature vector as input, whether the network flow data is associated with a network intrusion.

### Claim 11 (depends on 10)

The non-transitory machine-readable medium of claim 10, wherein the network intrusion includes a botnet, denial of service (DoS), distributed DoS (DDoS), brute force, structured query language (SQL) injection, infiltration, or port scan.

### Claim 12 (depends on 10)

The non-transitory machine-readable medium of claim 10, wherein the classifier is a clustering classifier that clusters the augmented feature vectors into groups including a first group corresponding to a benign network flow data and a second group corresponding to a malicious network flow data.

### Claim 13 (depends on 10)

The non-transitory machine-readable medium of claim 10, wherein the classifier performs k-means clustering or k-nearest neighbors (NN) clustering.

### Claim 14 (depends on 10)

The non-transitory machine-readable medium of claim 10, wherein the classifier is a multilayer perceptron (MLP).

### Claim 15

A method for network intrusion detection, the method comprising:
receiving a network flow data;
implementing a topological algorithm to identify respective (birth, death) pairs to which the network flow data maps;
appending the respective (birth, death) pairs to a feature vector representing the network flow data resulting in an augmented feature vector; and
determining, using a classifier or clustering algorithm that operates using the augmented feature vector as input, whether the network flow data is associated with a network intrusion.

### Claim 16 (depends on 15)

The method of claim 15, wherein data of the network flow data includes three or more of an input interface index, an output interface index, timestamps for flow start and finish times, number of bytes and packets observed in a corresponding flow, source and destination internet protocol (IP) address, internet control message protocol (ICMP) type and code, IP protocol, type of service value, source and destination port numbers for transmission control protocol (TCP), unigram data protocol (UDP), or stream control transmission protocol (SCTP), union of all TCP flags observed over a life of the flow, IP address of an immediate next-hop along a route to a destination, or source and destination IP masks.

### Claim 17 (depends on 16)

The method of claim 16, wherein the operations further comprise:
extracting features of the network flow data; and
wherein the topological algorithm operates on the extracted features to generate a persistence diagram for the network flow data along with other network flow data.

### Claim 18 (depends on 17)

The method of claim 17, further comprising, before extracting the features, completing the network flow data to alter text to numerical values and fill-in any empty fields of the network flow data.

### Claim 19 (depends on 18)

The method of claim 18, further comprising, after completing the network flow data, balancing a number of samples of each class in network flow data including the network flow data.

### Claim 20 (depends on 15)

The method of claim 15, wherein the network intrusion includes a botnet, denial of service (DoS), distributed DoS (DDoS), brute force, structured query language (SQL) injection, infiltration, or port scan.
