Controller area network decoder (CAN-D)

Patent metadata

Field	Value
Publication number	US-11780389-B2
Application number	US-202017117535-A
Country	US
Kind code	B2
Filing date	Dec 10, 2020
Priority date	Dec 13, 2019
Publication date	Oct 10, 2023
Grant date	Oct 10, 2023

Abstract

A system and method for decoding an unknown automotive controller area network (“CAN”) message definitions. CAN data vehicle signal mappings are typically held in secret and varied by automotive model and year. Without knowledge of the mappings, the wealth of real-time vehicle data hidden in the automotive CAN packets is uninterpretable—impeding research, after-market tuning, efficiency and performance monitoring, fault diagnosis, and privacy-related technologies. This technology can ascertain the CAN signals' boundaries (start bit and length), endianness (byte ordering), signedness (binary-to-integer encoding) from raw CAN data. This allows conversion of CAN data to time series. Interpreting the translated CAN data's physical meaning and finding a linear mapping to standard units (e.g., knowing the signal is speed and scaling values to represent units of miles per hour) can be achieved for many signals by leveraging diagnostic standards to obtain real-time measurements of in-vehicle systems. The system and method can be integrated into lightweight hardware enabling an OBD-II plugin for real-time in-vehicle CAN decoding or run on standard computers. The system can output a standard DBC file with the signal definition information.

First claim

The embodiments of the invention in which an exclusive property or privilege is claimed are defined as follows:

1. A data-driven, vehicle-agnostic method for reverse engineering controller-area network (CAN) message definitions, the method comprising:
receiving a plurality of CAN packets from a CAN of a vehicle, each CAN packet comprising a message, the CAN message packing signals encoded based on unknown definitions established by third parties, each signal definition including information on how to tokenize, translate, and interpret the signal, where the information to tokenize is configured to demarcate a sequence of bits corresponding to the signal and includes start bit, bit-sequence length, and byte ordering, where the information to translate is configured to convert the sequence of bits to integers by indicating whether unsigned or signed binary-to-integer encoding was used to encode the sequence of bits, and
where the information to interpret comprises label and unit giving physical meaning of the signal and its units, and a scale and offset providing a linear mapping of the signal to its units;
for each CAN packet of the received plurality of CAN packets, partitioning the message into signals having defined but unknown start bits, lengths, byte order, and signed or unsigned encoding to obtain a respective time series of integers for each of the signals, wherein partitioning the message into signals comprises learning signal boundary probabilities by at least one of an unsupervised signal boundary classification algorithm and a supervised signal boundary classification algorithm, then optimizing, based on the signal boundary probabilities, the signal packing likelihood within the message to tokenize signals by identifying the byte orderings within the tokenized signals, and
classifying each tokenized signal as signed or unsigned by at least one of a supervised learning method or an unsupervised learning method;
converting each tokenized signal to a time series of integers based on signedness classification;
receiving time series of diagnostic responses corresponding to physical measurements in the vehicle; and
interpreting at least some of the time series of integers as corresponding physical measurements by comparing each time series of integers to a plurality of time series of labeled data to match the time series of integers to a corresponding one of the time series of labeled data.

2. The method of claim 1 , wherein the plurality of time series of labeled data includes one or more of timestamped diagnostic responses, timestamped GPS data, timestamped yaw, pitch, and roll data from an accelerometer disposed in the vehicle, and timestamped yaw, pitch, and roll data from an inertial measurement unit disposed in the vehicle.

3. The method of claim 1 , wherein the plurality of time series of labeled data includes a plurality of time series of diagnostic responses and wherein the comparing includes determining a linear mapping between the time series of integers and a particular time series of diagnostic response.

4. The method of claim 1 , wherein the physical measurements in the vehicle comprise measurements available via Unified Diagnostic Standard.

5. An apparatus for real-time interpretation of vehicle controller area network (CAN) data, the apparatus comprising:
a vehicle CAN interface configured to receive vehicle CAN frames, each vehicle CAN frame having an identifier (ID) and CAN data, wherein the CAN data is encoded based on an unknown signal definition established by a third party, each unknown signal definition including information to tokenize, translate, and interpret the CAN data, where the information to tokenize includes information to demarcate sequences of bits corresponding to signals in the CAN data and byte ordering, wherein the information to translate includes information about how the sequences of bits were converted to integers;
memory configured to store an ID trace including CAN data from different vehicle CAN frames associated with a particular vehicle CAN frame ID;
a processor configured to generate a vehicle CAN signal definition for CAN data associated with the particular vehicle CAN frame ID based on the ID trace, wherein the processor is configured to predict signal boundaries within the CAN data based on the CAN data in the ID trace and generate signal boundary probabilities, wherein the processor is configured to predict endianness of signals within the CAN data based on the signal boundary probabilities and generate tokenized signals, wherein the processor is configured to predict signedness of the tokenized signals and generate translated signals, wherein the processor is configured to match the translated signals to external labeled timeseries and generate interpreted signals, wherein the processor is configured to generate the vehicle CAN signal definition based on the interpreted signals and store the generated signal definition in memory; and
wherein the processor is configured to decode CAN data in CAN data frames with the particular vehicle CAN frame ID received by the vehicle CAN interface according to the signal definition stored in memory.

6. The apparatus of claim 5 , wherein the CAN data includes up to 64 bits of data.

7. The apparatus of claim 5 , wherein the signal boundary probabilities include bit-wise probabilities of bit gaps between consecutive bits in the CAN data.

8. The apparatus of claim 5 , wherein the signal boundary probabilities include a first set of signal boundary probabilities that assume big endian byte ordering and a second set of signal boundary probabilities that assume little endian byte ordering, wherein the processor is configured to predict endianness of signals within the CAN data using an optimization algorithm based on the first set of signal boundary probabilities and the second set of signal boundary probabilities.

9. The apparatus of claim 5 , wherein the processor is configured to predict the likelihood of signal boundaries within the CAN data as a function of bits local to CAN data bit gaps.

10. The apparatus of claim 9 wherein the processor is configured to predict the likelihood of signal boundaries within the CAN data utilizing a heuristic.

11. The apparatus of claim 5 , wherein the processor is configured to generate signal boundary probabilities based on a trained supervised machine learning signal boundary model.

12. The apparatus of claim 9 , wherein the processor is configured to predict endianness of signals within the CAN data based on signal boundary probabilities generated as a function of bits local to CAN data bit gaps and a plurality of different assumed endianness es.

13. The apparatus of claim 5 , wherein the processor is configured to predict endianness of signals within the CAN data based on a cost function that balances partitioning CAN data into too many signals and joining multiple disparate signals.

14. The apparatus of claim 13 , wherein the balance is struck by balancing a cut penalty β with a join penalty f.

15. The apparatus of claim 5 , wherein the processor is configured to predict signedness of the tokenized signals and generate translated signals with a CAN-D signedness heuristic.

16. The apparatus of claim 15 , wherein the CAN-D signedness heuristic is based on how the two most significant bits of a signal behave when the signal is signed and unsigned.

Assignees

• Ut Battelle Llc

Inventors

• Verma Kiren E
• Bridges Robert A
• Hollifield Samuel C

Classifications

• B60R16/0232 Primary

  for measuring vehicle parameters and indicating critical, abnormal or dangerous conditions · CPC title

• G06N20/00 Primary

  Machine learning · CPC title

• G07C5/008

  communicating information to a remotely located station (transmission systems for measured values G08C) · CPC title

• G07C5/0808

  Diagnosing performance data (testing of vehicles G01M17/00; testing of electrical installation on vehicles G01R31/005) · CPC title

• G07C5/10

  using counting means or digital clocks · CPC title