Inferring building metadata from distributed sensors

• US 9,240,111 B2
• Filed: 10/06/2010
• Issued: 01/19/2016
• Est. Priority Date: 10/06/2010
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation), Samsung Electro-Mechanics Co., Ltd. (Samsung Group)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Scott, James W., Constandache, Ionut, Hodges, Stephen Edward
• Primary Examiner(s): Lau, Hoi
• Application Number:
  US12/899,204
• Publication Number:
  US 20120086568A1
• Time in Patent Office: 1,931 Days
• Field of Search: 340/501, 340/566, 340/505, 340/539.26, 340/540, 340/541, 348/148, 348/500, 348/164, 348/461, 348/218.1, 348/584, 348/E5.024, 348/E9.055, 370/503, 455/456.1
• US Class Current: 1/1
• CPC Class Codes: G05B 15/02 electric
• CPC Class Codes: G08B 13/00 Burglar, theft or intruder alarms
• CPC Class Codes: G08B 21/22 responsive to presence or absence of persons

Abstract

Inferring building metadata from distributed sensors is described. In an embodiment multiple sensors are situated at various locations in a building detecting physical quantities (e.g. light, heat, motion). The data from at least one sensor is analyzed in order to detect events occurring in proximity to the sensor. Data about an event detected at a first sensor is compared with events detected at other sensors in order to identify correlated events from which connections between the sensor locations can be inferred and a building layout generated. In some embodiments a threshold may be applied to the data in order to filter out false events. In some embodiments the building layout may be used as part of a building control system. In some embodiments the sensor data may be used to determine if the sensor has been moved.

Claims

1. A computer implemented method comprising:

accessing sensor data from one of a plurality of distributed sensors, each sensor located in a respective building portion of a building;

analyzing the sensor data to identify events indicated by a change in the sensor data;

comparing events in sensor data from at least two of the plurality of distributed sensors to identify correlated events and for each identified correlated event, storing data representing a link between corresponding sensors; and

analyzing the stored link data to automatically generate metadata describing a physical arrangement of the respective building portions, each respective building portion associated with at least one sensor and the physical arrangement based at least in part on the data representing links between corresponding sensors.

2. A method according to claim 1, further comprising:

filtering the stored link data, prior to analyzing the data, by applying a threshold condition based on a number of links between pairs of sensors.

3. A method according to claim 2, wherein filtering the stored link data comprises:

selecting a first and a second sensor from the plurality of distributed sensors;

computing a ratio of a number of links between the first sensor to the second sensor divided by a total number of links involving a selected one of the first sensor and the second sensor; and

excluding links between the first and second sensors from the stored link data where the ratio does not exceed a predefined value.

4. A method according to claim 1, further comprising:

controlling a building system based at least on the generated metadata.

5. A method according to claim 4, wherein the building system is controlled based at least on the generated metadata and recent sensor data.

6. A method according to claim 4, wherein the building system comprises one of a temperature control system and a security system.

7. A method according to claim 1, wherein the metadata comprises one of a room adjacency graph and an auto-generated floor plan.

8. A method according to claim 1, wherein comparing events in sensor data from at least two of the plurality of distributed sensors to identify correlated events comprises identifying events in sensor data from at least two of the plurality of distributed sensors which occur within a predefined time interval.

9. A method according to claim 1, wherein the plurality of distributed sensors comprises sensors of a first type and wherein the method further comprises:

accessing sensor data from one of a second plurality of distributed sensors, each sensor located in a portion of a building and comprising a sensor of a second type;

analyzing the sensor data to identify events indicated by a change in the sensor data; and

comparing events in sensor data from at least two of the second plurality of distributed sensors to identify correlated events based at least on a time-synchronization of event and for each identified correlated event, storing data representing a link between corresponding sensors; and

combining the stored link data from each of the pluralities of distributed sensors prior to analyzing the stored link data.

10. A method according to claim 9, further comprising one of:

filtering the stored link data from each of the pluralities of distributed sensors by applying a threshold condition based on a number of links between pairs of sensors, prior to combining the stored link data; and

filtering the combined stored link data from each of the pluralities of distributed sensors by applying a threshold condition based on a number of links between pairs of sensors, prior to analyzing the stored link data.

11. A method according to claim 1, further comprising:

analyzing the sensor data to detect relocation of a sensor; and

updating the metadata.

12. A building control system comprising:

an input arranged to receive sensor data from a plurality of sensors distributed among respective building portions within a building;

an event detection engine adapted to analyze the sensor data and record events for each sensor corresponding to changes in the sensor data;

a comparison engine adapted to compare time-stamped events for different sensors to identify correlated events and generate data representing a transition between sensors for each identified correlated event; and

an analysis engine adapted to automatically generate a model defining the relative spatial positions of the respective building portions based on the transition data each respective building portion associated with at least one sensor and the relative spatial positions being based at least in part on the data representing transitions between sensors for the identified correlated events.

13. A system according to claim 12, further comprising:

a filtering engine adapted to filter the transition data, prior to input to the analysis engine, based on a number of transitions between pairs of sensors.

14. A system according to claim 13, wherein the filtering engine is arranged to filter the transition data based on the number of transitions between pairs of sensors and based on a total number of transitions associated with one of the pair of sensors.

15. A system according to claim 12, further comprising:

an output adapted to send a control signal to an element in one of a building temperature control system and a building security system based on the model and current sensor data.

16. A system according to claim 12, wherein the analysis engine is further adapted to analyze the sensor data to determine occupancy of a building portion and to predict future occupancy of the building portion based on the determined occupancy;

and the system further comprises;

an output adapted to send a control signal to a temperature control element associated with a building portion based on the model and predicted future room occupancy.

17. A system according to claim 12, wherein the plurality of sensors comprise motion sensors and the model defining the relative spatial positions of the building portions comprises a set of possible transitions between building portions and wherein the analysis engine is further adapted to analyze the sensor data in real-time to identify any transitions outside the set of possible transitions;

and

to raise an alarm on identification of a transition outside the set of possible transitions.

18. One or more tangible device-readable media with device-executable instructions that, when executed by a computing system, direct the computing system to control a building system by performing steps comprising:

receiving sensor data from one of a plurality of distributed sensors of a first type, each sensor located in a respective building portion of a building;

analysing the sensor data to identify events indicated by a change in the sensor data;

comparing events in sensor data from at least two of the plurality of distributed sensors of the first type to identify events and for each identified event, storing data representing a transition between corresponding sensors;

filtering the data representing transitions between sensors using a threshold to identify connections between building portions;

analyzing the connections to automatically generate a representation of the spatial relationship between respective building portions, each respective building portion being associated with at least one sensor and the spatial relationship between building portions being based at least in part on the filtered data representing transitions between sensors; and

generating a control signal for the building system based on the spatial relationship and current sensor data.

19. One or more tangible device-readable media according to claim 18, further comprising device-executable instructions that, when executed by a computing system, direct the computing system to perform steps comprising:

receiving sensor data from one of a plurality of distributed sensors of a second type, each sensor located in a portion of a building;

analyzing the sensor data to identify events indicated by a change in the sensor data; and

comparing events in sensor data from at least two of the plurality of distributed sensors of the second type to identify time synchronized events and for each identified time synchronized event, storing data representing a transition between corresponding sensors;

and wherein at least one of the filtering and analyzing steps uses transition or connection data generated using sensors of the first type and sensors of the second type.

20. One or more tangible device-readable media according to claim 19, wherein the sensors of a first type comprise PIR sensors and the sensors of the second type comprise light sensors.