Method and systems using privacy-preserving analytics for aggregate data

• US 10,216,959 B2
• Filed: 08/01/2016
• Issued: 02/26/2019
• Est. Priority Date: 08/01/2016
• Status: Active Grant

• Current Assignee: Mitsubishi Electric Corporation, Mitsubishi Electric Research Laboratories, Inc. (Mitsubishi Electric Corporation)
• Original Assignee: Mitsubishi Electric Corporation, Mitsubishi Electric Research Laboratories, Inc. (Mitsubishi Electric Corporation)
• Inventors: Wang, Ye, Hattori, Mitsuhiro, Hirano, Takato, Shimizu, Rina, Matsuda, Nori
• Primary Examiner(s): Gee, Jason K
• Assistant Examiner(s): Cattungal, Dereena T
• Application Number:
  US15/224,742
• Publication Number:
  US 20180032758A1
• Time in Patent Office: 939 Days
• Field of Search: 726 30
• CPC Class Codes: G06F 21/6254 by anonymising data, e.g. decorrelating personal data from the owner's identification
• CPC Class Codes: G06Q 50/06 Energy or water supply
• CPC Class Codes: H04L 63/0421 Anonymous communication, i.e. the party's identifiers are hidden from the other party or parties, e.g. using an anonymizer

Abstract

Methods and systems for transmitting user aggregate data to a third party, such that a privacy of the aggregated data is protected, while analytical usefulness of the aggregated data is preserved. The method including receiving, using a transceiver, aggregated data including time-series data collected over a period of time. Selecting, from a memory, a mapping for transforming a segment of the aggregated data of a predetermined size. Partitioning the aggregated data into a multiple data segments, each data segment is of the predetermined size. Transforming each data segment using the mapping to produce multiple transformed data segments, wherein each data segment is transformed by the mapping independently from other data segments. Finally, transmitting, using the transceiver, the multiple transformed data segments to a third party over a communication channel, wherein steps of the method are performed by a processor operatively connected with the memory and the transceiver.

Claims

1. A method for transmitting aggregated data to a third party, such that a privacy of the aggregated data is protected, while analytical usefulness of the aggregated data is preserved, comprising:

receiving, using a transceiver, aggregated data including time-series data collected over a period of time;

selecting, from a memory, a mapping for transforming a segment of the aggregated data of a predetermined size;

partitioning the aggregated data into a multiple data segments, each data segment is of the predetermined size;

transforming each data segment using the mapping to produce multiple transformed data segments, wherein each data segment is transformed by the mapping independently from other data segments, and each mapped data segment of the aggregated data modifies the data segment such that the privacy of the data segment is protected, while analytical usefulness of the data segment is preserved;

and transmitting, using the transceiver, the multiple transformed data segments to a third party over a communication channel, wherein steps of the method are performed by a processor operatively connected with the memory and the transceiver;

wherein the mapping further comprises;

collecting a training set of the aggregated data;

determining a statistical model fitting the training set of the aggregated data; and

determining the mapping using the statistical model by optimizing a balance between a criteria for protecting privacy of the transformed aggregated data and a criteria of analytical usefulness of the transformed aggregated data, wherein the criteria of analytical usefulness is above a threshold criteria to allow for third party analytics of the transformed aggregated data by the third party.

2. The method of claim 1, further comprising:

combining the multiple transformed data segments to form transformed aggregated data; and

transmitting the transformed aggregated data to the third party.

3. The method of claim 1, further comprising:

updating the statistical model based on the received aggregated data; and

updating the mapping based on the updated statistical model.

4. The method of claim 1, wherein the statistical model is a Factorial Hidden Markov Model (FHMM).

5. The method of claim 1, further comprising:

selecting the mapping from a set of mappings stored in the memory using a property of the received aggregated data, wherein each mapping from the set of mapping is determined for a different steady state of the aggregated data.

6. The method of claim 5, wherein the property of the received aggregated data is one of:

a time property of the period of time including one or combination of a length of the period of time, a timing of the period of time and a season of the period of time, and

a time property related to a behavioral use property considered as privacy data of the period of time.

7. The method of claim 5, further comprising:

retrieving from the memory a set of statistical models, each model corresponds to the mapping from the set of mappings;

fitting the received aggregated data into each statistical model to determining a fitting error for each statistical model; and

selecting the mapping from the set of mapping that corresponds to the statistical model with the smallest fitting error.

8. The method of claim 1, wherein the aggregate data includes usage of at least one power consumer by a user, wherein the at least one power consumer includes a structure, a part of a structure, an electronic device, a power consuming device or some combination thereof.

9. The method of claim 1, wherein the multiple data segments of the aggregate data refers to an ongoing and/or future power consumption event by a user having at least one power consumer.

10. The method of claim 1, wherein the aggregate data originates from a user, and is user energy data collected using a metering device located on a consumer side of the metering device.

11. The method of claim 1, wherein the aggregate data originating from a user is energy data collected by the user using a user metering device or a user measuring device located on a user side of a third party metering device, such that the user transmits, using a user transmitter, the transformed aggregated data to the third party over a user communication channel, wherein steps of the method are performed by the processor operatively connected with the user transmitter.

12. A method for transmitting aggregated data to a third party, such that a privacy of the aggregated data is protected, while analytical usefulness of the aggregated data is preserved, comprising:

receiving, using a transceiver, aggregated data including time-series data collected over a period of time;

selecting, from a memory, a mapping for transforming a segment of the aggregated data of a predetermined size, wherein the selecting of the mapping is from a set of mappings stored in the memory using a property of the received aggregated data, wherein each mapping from the set of mapping is determined for a different steady state of the aggregated data;

partitioning the aggregated data into a multiple data segments, each data segment is of the predetermined size;

transforming each data segment using the mapping to produce multiple transformed data segments, wherein each data segment is transformed by the mapping independently from other data segments, and each mapped data segment of the aggregated data modifies the data segment such that the privacy of the data segment is protected, while analytical usefulness of the data segment is preserved;

and combining the multiple transformed data segments to form transformed aggregated data; and

transmitting, using the transceiver, the transformed aggregated data to a third party over a communication channel, wherein steps of the method are performed by a processor operatively connected with the memory and the transceiver;

wherein the mapping further comprises;

collecting a training set of the aggregated data;

determining a statistical model fitting the training set of the aggregated data; and

determining the mapping using the statistical model by optimizing a balance between a criteria for protecting privacy of the transformed aggregated data and a criteria of analytical usefulness of the transformed aggregated data, wherein the criteria of analytical usefulness is above a threshold criteria to allow for third party analytics of the transformed aggregated data by the third party.

13. The method of claim 12, wherein the property of the received aggregated data is one of:

a time property of the period of time including one or combination of a length of the period of time, a timing of the period of time and a season of the period of time; and

a time property related to a behavioral use property considered as privacy data of the period of time.

14. A system for transmitting aggregated data to a third party, such that a privacy of the aggregated data is protected, while analytical usefulness of the aggregated data is preserved, comprising:

a transceiver for receiving aggregated data including time-series data collected over a period of time over a communication channel;

a memory to store at least one mapping for transforming a segment of the aggregated data of a predetermined size;

a processor operatively connected with the memory and the transceiver, the process is configured for;

selecting, from the memory, a mapping for transforming a segment of the aggregated data of a predetermined size from the at least one mapping, partitioning the aggregated data into a multiple data segments, each data segment is of the predetermined size, transforming each data segment using the mapping to produce multiple transformed data segments, wherein each data segment is transformed by the mapping independently from other data segments, and each mapped data segment of the aggregated data modifies the data segment such that the privacy of the data segment is protected, while analytical usefulness of the data segment is preserved;

and transmitting, using the transceiver, the multiple transformed data segments to a third party over a communication channel;

wherein the mapping further comprises;

collecting a training set of the aggregated data;

determining a statistical model fitting the training set of the aggregated data; and

determining the mapping using the statistical model by optimizing a balance between a criteria for protecting privacy of the transformed aggregated data and a criteria of analytical usefulness of the transformed aggregated data, wherein the criteria of analytical usefulness is above a threshold criteria to allow for third party analytics of the transformed aggregated data by the third party.

15. The system of claim 14, wherein the processor is further configured for:

selecting the mapping from a set of mappings stored in the memory using a property of the received aggregated data, wherein each mapping from the set of mapping is determined for a different steady state of the aggregated data,

wherein the property of the received aggregated data is one of;

a time property of the period of time including one or combination of a length of the period of time, a timing of the period of time and a season of the period of time, and a time property related to a behavioral use property considered as privacy data of the period of time.

16. The system of claim 15, wherein the processor is further configured for:

retrieving from the memory a set of statistical models, each model corresponds to the mapping from the set of mappings;

fitting the received aggregated data into each statistical model to determining a fitting error for each statistical model; and

selecting the mapping from the set of mapping that corresponds to the statistical model with the smallest fitting error.