Granted Patent B2

US 12,003,256 · App. 18/460,553 · Granted Jun 4, 2024

Inventors: Joshua Cooper (Columbia, SC); Aliasghar Riahi (Orinda, CA); Charles Yeomans (Orinda, CA)

Assignee: ATOMBEAM TECHNOLOGIES INC.

H03M7/3059G06F21/554G06N20/00H03M7/6005

Abstract

Claims (55)

1. A system for data compaction with intrusion detection, comprising:

a computing device comprising a processor and a memory;

an intrusion detection module comprising a first plurality of programming instructions stored in the memory which, when operating on the processor, causes the computing device to:

receive a codeword data stream;

use one or more algorithms to compute a probability distribution of a plurality of codewords within the codeword data stream;

compare the computed probability distribution with a reference probability distribution to compute an amount of divergence between the computed probability distribution and the reference probability distribution; and

when the computed amount of divergence exceeds a configured risk sensitivity threshold:

store the computed divergence, the computed probability distribution, and the codeword as anomalous event data in a database;

generate an intrusion alert, the intrusion alert comprising the anomalous event data; and

send the intrusion alert to a user interface to be viewed by a user; and

a codebook training module comprising a second plurality of programming instructions stored in the memory which, when operating on the processor, causes the computing device to:

receive a training dataset;

use the training dataset to create the reference probability distribution;

send the reference probability distribution to the intrusion detection module;

receive data;

format the received data into a test dataset;

retrieve a first measured probability distribution associated with the previous training dataset from a monitor database;

use one or more algorithms to measure a second probability distribution of the test dataset;

compare the first and second measured probability distributions to compute the difference in distribution statistics between the test dataset and the previous training dataset;

in response to the difference in distribution statistics' exceeding a pre-determined difference threshold:

use the test dataset to retrain encoding and decoding algorithms;

utilize the retrained algorithms to create new data sourceblocks;

create new codeword for each new data sourceblock;

store each new data sourceblock and its associated new codeword in an updated codebook; and

send the updated codebook to a plurality of encoding and decoding machines.

2. The system of claim 1 , wherein the monitor database is stored in the memory of the computing device, wherein the monitor database comprises a previous training dataset, the first-measured probability distribution associated with the previous training dataset, performance metrics, and model predictions.

3. The system of claim 1 , further comprising a data deconstruction engine comprising a third plurality of programming instructions stored in the memory which, when operating on the processor, causes the computing device to:

receive a plurality of codewords from a codeword storage; and

send the plurality of codewords as a codeword data stream to the intrusion detection module.

4. A method for data compaction with intrusion detection, comprising the steps of:

receiving a codeword data stream;

using one or more algorithms to compute a probability distribution of a plurality of codewords within the codeword data stream;

comparing the computed probability distribution with a reference probability distribution to compute an amount of divergence between the computed probability distribution and the reference probability distribution; and

when the computed amount of divergence exceeds a configured risk sensitivity threshold:

storing the computed divergence, the computed probability distribution, and the codeword as anomalous event data in a database;

generating an intrusion alert, the intrusion alert comprising the anomalous event data; and

sending the intrusion alert to a user interface to be viewed by a user;

receiving a training dataset;

using the training dataset to create the reference probability distribution;

sending the reference probability distribution to the intrusion detection module;

receiving data;

formatting the received data into a test dataset;

retrieving a first measured probability distribution associated with the previous training dataset from a monitor database;

using one or more algorithms to measure a second probability distribution of the test dataset;

comparing the first and second measured probability distributions to compute the difference in distribution statistics between the test dataset and the previous training dataset;

in response to the difference in distribution statistics' exceeding a pre-determined difference threshold:

using the test dataset to retrain encoding and decoding algorithms;

utilizing the retrained algorithms to create new data sourceblocks;

creating new codeword for each new data sourceblock;

storing each new data sourceblock and its associated new codeword in an updated codebook; and

sending the updated codebook to a plurality of encoding and decoding machines.

5. The method of claim 4 , wherein the monitor database is stored in the memory of the computing device, wherein the monitor database comprises a previous training dataset, the first-measured probability distribution associated with the previous training dataset, performance metrics, and model predictions.

6. The method of claim 4 , further the steps of:

receiving a plurality of codewords from a codeword storage; and

sending the plurality of codewords as a codeword data stream to the intrusion detection module.

Assignments (1)

Continuity (20)