# US 2020/0267171 A1 · Systems and methods for detecting a communication anomaly

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

- Number: US 2020/0267171 A1
- Document ID: US20200267171A1
- Status: Published application
- Kind: A1
- Application number: 16/279,591
- Filed: 2019-02-19
- Published: 2020-08-20
- Earliest claimed: 2019-02-19
- Assignees: The Aerospace Corporation
- Applicants: The Aerospace Corporation
- Inventors: Mohammad Mozumdar; Philip A. Dafesh
- Web page: https://patentgrep.com/patent/US20200267171A1

## CPC

- H04L 63/1425: {Traffic logging, e.g. anomaly detection}
- G06F 13/4282: {on a serial bus, e.g. I2C bus, SPI bus (on daisy chain buses G06F13/4247)}
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/00: Machine learning
- H04L 12/40013: {Details regarding a bus controller}
- H04L 63/1416: {Event detection, e.g. attack signature detection}
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Cyberattacks are rampant and can play a major role in modern warfare, particularly on a widely adopted platforms such as the MIL-STD-1553 standard. To protect a 1553 communication bus system from attacks, a trained statistical or machine learning model can be used to monitor commands from a bus controller of the 1553 communication bus system. The statistical and/or machine learning model can be trained to recognize communication anomalies based at least on the probability distribution of patterns of one or more commands. The statistical model can be stochastic model such as a Markov chain that describes a sequence of possible commands in which the probability of each command depends on the occurrence of a group of one or more commands.

## Claims

### Claim 1

A method for detecting a communication anomaly on a communication bus, the method comprising:
segmenting a training data set into a plurality of segments, wherein the training data set comprises of commands from a bus controller of the communication bus;
identifying patterns within each of the plurality of segments, wherein each pattern comprises one or more commands; and
generating a statistical model representing probability relationships between identified patterns, wherein the probability relationships define the probability of a first pattern to occur after a second pattern.

### Claim 2 (depends on 1)

The method of claim 1, wherein segmenting the training data set comprises segmenting the training data at every time interval, between two commands, where it is above a predetermined time interval threshold.

### Claim 3 (depends on 2)

The method of claim 2, wherein the predetermined time interval threshold comprises a mean value of time intervals between every two sequential commands of the training data set.

### Claim 4 (depends on 3)

The method of claim 3, wherein the mean value comprises one of a winsorized mean, a truncated mean, or a modified mean value.

### Claim 5 (depends on 1)

The method of claim 1, wherein the statistical model comprises a Markov chain model.

### Claim 6 (depends on 1)

The method of claim 1, wherein identifying patterns within each of the plurality of segments comprises identifying largest possible patterns before identifying smaller patterns within each segment.

### Claim 7 (depends on 1)

The method of claim 1, wherein identifying patterns within each of the plurality of segments comprises identifying one or more commands that do not fit into any pattern.

### Claim 8 (depends on 1)

The method of claim 1, wherein the patterns are non-overlapping.

### Claim 9 (depends on 1)

The method of claim 1, wherein the communication bus comprises a MIL-STD-1533 or controller area network (CAN) communication bus.

### Claim 10 (depends on 1)

The method of claim 1, further comprising:
monitoring commands transmit from a second bus controller to one or more remote terminals; and
determining whether a first command in a group of commands from the second bus controller to the one or more remote terminals is abnormal using the statistical model.

### Claim 11

A system for detecting a communication anomaly on a communication bus, the system comprising:
a memory;
one or more processors coupled to the memory, the one or more processors configured to:
segment a training data set into a plurality of segments, wherein the training data set comprises of commands from a bus controller of a communication bus;
identify patterns within each of the plurality of segments, wherein each pattern comprises one or more commands; and
generate a Markov chain model representing probability relationships between identified patterns, wherein the probability relationships define the probability of a first pattern to occur after a second pattern.

### Claim 12 (depends on 11)

The system of claim 11, wherein the one or more processors are further configured to segment the training data set at every time interval, between two commands, where it is above a predetermined time interval threshold.

### Claim 13 (depends on 12)

The system of claim 12, wherein the predetermined time interval threshold comprises a mean value of time intervals between every two sequential commands of the training data set.

### Claim 14 (depends on 13)

The system of claim 13, wherein the mean value comprises one of a winsorized mean, a truncated mean, or a modified mean value.

### Claim 15 (depends on 11)

The system of claim 11, wherein the one or more processors are configured to determine whether the first command is abnormal based at least on a probability value of a transition between one or more preceding groups of command and the first command.

### Claim 16 (depends on 11)

The system of claim 11, wherein the one or more processors are configured to identify patterns within each of the plurality of segments by identifying largest possible patterns before identifying smaller patterns within each segment.

### Claim 17 (depends on 16)

The system of claim 16, wherein the one or more processors are configured to identify one or more commands within a segment that do not fit into any pattern.

### Claim 18 (depends on 11)

The system of claim 11, wherein the one or more processors are configured to identify patterns that are non-overlapping.

### Claim 19 (depends on 11)

The system of claim 11, wherein the one or more processors are configured to:
monitor commands transmit from a second bus controller to one or more remote terminals; and
determine whether a first command in a group of commands from the second bus controller to the one or more remote terminals is abnormal using the generated Markov chain model.

### Claim 20

A method for detecting a communication anomaly on a communication bus, the method comprising:
monitoring commands transmit from a bus controller to one or more remote terminals, wherein the bus controller and the one or more remote terminals are coupled to the communication bus; and
determining whether a first command in a group of commands from the bus controller to the one or more remote terminals is abnormal using a trained Markov chain model that defines a probability of the first command occurring after a group of one or more commands previously sent by the bus controller, wherein the Markov chain model is trained using identified patterns in each of a plurality of training data segments, wherein training data segments are segmented based on a statistical analysis of time intervals between two consecutive commands of an entire training data set.
