Traffic and geometry modeling with sensor networks

• US 7,302,369 B2
• Filed: 10/10/2003
• Issued: 11/27/2007
• Est. Priority Date: 10/10/2003
• Status: Active Grant

• Current Assignee: Mitsubishi Electric Research Laboratories, Inc. (Mitsubishi Electric Corporation)
• Original Assignee: Mitsubishi Electric Research Laboratories, Inc. (Mitsubishi Electric Corporation)
• Inventors: Wren, Christopher R.
• Primary Examiner(s): Rodriguez; Paul
• Assistant Examiner(s): Ochoa; Juan Carlos
• Application Number:
  US10/684,116
• Publication Number:
  US 20050080601A1
• Time in Patent Office: 1,509 Days
• Field of Search: 703/2, 701/117
• US Class Current: 703/2
• CPC Class Codes: G05B 15/02 electric

Abstract

A method models movement of users in an environment including sensors connected in a network. Events due to movement of the users are detected at the sensors and each event is labeled according to a particular sensor and time of the event. The events for each sensor are summed into a corresponding histogram time interval bin. A plurality of co-occurrence matrices are generated from the histograms according to

$C_{i,j,\delta }=\sum _{t=0}^T{H}_{i,t}{H}_{j,t+\delta },$
where i and j represent each possible pair of sensors, δ represent time-off-sets, T is a total time for the detecting, t represents a particular time, and H represents the histogram time interval bins. The co-occurrence matrices can be used to determine a geometry of the network, and for predicting future activities signaled by terminating events.

Claims

1. A method for modeling movement of users in an environment, the environment including sensors connected in a network, comprising:

detecting events due to movement of the users at the sensors;

labeling each event according to a particular sensor that detected the event and time of the event;

summing the events for each sensor into a corresponding histogram time interval bin associated with the sensor;

generating a plurality of co-occurrence matrices from the histograms according to

$C_{i,j,\mathrm{\delta <br><br>}}=\underset{t=0}{\overset{T}{\sum <br><br>}}<br><br><br><br>H_{i,t}<br><br>H_{j,t+\mathrm{\delta <br><br>}},$

 

where i and j represent each possible pair of sensors, δ

represent time off-sets, T is a total time for the detecting, t represents a particular time, and H represents the histogram time interval bins.

2. The method of claim 1, in which the environment is inside a building.

3. The method of claim 1, in which the environment is an open space.

4. The method of claim 1, in which the network is wired.

5. The method of claim 1, in which the network is wireless.

6. The method of claim 1, in which the network includes wired and wireless sensors.

7. The method of claim 1, in which the sensors have a resolution of one bit to detect the events.

8. The method of claim 1, in which a sensing modality is selected from the group consisting of infra-red, thermal, ultrasonic, light, radar, sonar, microwave, pressure, on-off switch, and combinations thereof.

9. The method of claim 1, in which the sensors are globally distributed in the environment in an ad-hoc arbitrary pattern.

10. The method of claim 1, in which ranges of some sensors are non-overlapping.

11. The method of claim 1, further comprising:

determining, for each possible pair of sensors, a distance between the pair of sensors; and

mapping the distances to a geometry of the sensors.

12. The method of claim 11, in which the summed events include peaks, and each peak reflects a relative time-offset between a particular pair of sensors.

13. The method of claim 12 in which the determining further comprises:

applying a consistent average velocity of the movement of the users over time to the relative time-offset between the pair of sensors.

14. The method of claim 11, in which the mapping uses multi-dimensional scaling.

15. The method of claim 14, further comprising:

evaluating a likelihood that a particular geometry conforms to the estimated distances using a stress measure Φ

=Σ

└

d_ij−

f (δ

_ij)┘

, where f (ij) is a non-metric, monotone transformation.

16. The method of claim 15, in which the stress measure rank-orders the distances between the pairs of sensors.

17. The method of claim 1, in which a particular sensor detects events associated with predicting a future activity, and further comprising:

determining, from the co-occurrence matrices, distributions of probabilities that the terminating event occurs based on the events detected by all of the other sensors other than the particular sensor; and

summing the distributions of probabilities to predict the probability that the future activity will occur prior to detecting the terminating event while detecting the events.

18. The method of claim 17, in which there is one distribution for each other sensor.

19. A system for modeling movement of users in an environment, comprising:

a plurality of sensors distributed throughout the environment;

means for labeling events detected by the sensors according to a particular sensor that detected the event and time of the event;

means for summing the events for each sensor into a corresponding histogram time interval bin associated with the sensor; and

means for generating a plurality of co-occurrence matrices from the histograms according to

$C_{i,j,\mathrm{\delta <br><br>}}=\underset{t=0}{\overset{T}{\sum <br><br>}}<br><br><br><br>H_{i,t}<br><br>H_{j,t+\mathrm{\delta <br><br>}},$

 

where i and j represent each possible pair of sensors, δ

represent time off-sets, T is a total time for the detecting, t represents a particular time, and H represents the histogram time interval bins.

20. The system of claim 19, further comprising:

means for determining, for each possible pair of sensors, a distance between the pair of sensors; and

means for mapping the distances to a geometry of the sensors.

21. The system of claim 19, in which a particular sensor detects events associated with a future activity, and further comprising:

means for determining, from the co-occurrence matrices, distributions of probabilities that the terminating event occurs based on the events detected by all of the other sensors other than the particular sensor; and

means for summing the distributions of probabilities to predict the probability that the future activity will occur prior to detecting the terminating event while detecting the events.