Method and apparatus for using mobile code for distributed data fusion in networked sensing systems
View Patent ↗A sensing system includes a set of sensors and a data-fusing mechanism coupled to at least one of these sensors. In the set of sensors, at least one sensor is configured to store one or more measurement models for one or more phenomenon states. Furthermore, at least one sensor in the set of sensors is configured to sample a measurement value and generate a likelihood function based on the sampled measurement and the measurement models. The data-fusing mechanism coupled to a respective sensor in the set of sensors is configured to collect one or more likelihood functions generated by the one or more sensors and use the collected likelihood functions to compute an aggregate probability of a phenomenon state.
1. A sensing system, comprising:
a set of sensors, each storing at least one measurement model for detecting a phenomenon state, wherein a respective sensor samples a measurement value;
a data-fusing mechanism coupled to the sensor, which stores a first measurement model as P(Z=z|X=x),
wherein Z corresponds to the measurement value and X corresponds to a first phenomenon state under investigation, and
wherein the data-fusing mechanism collects likelihood functions generated by the set of sensors, and computes an aggregate probability of a second phenomenon state based on
∏
i
=
1
n
P
(
Z
i
=
z
i
❘
X
=
x
)
which corresponds to a product of a number of n likelihood collected from the sensors.
2. The sensing system of claim 1 , wherein the sensor is configured to communicate the likelihood function to the data-fusing mechanism in the form of a mobile code.
3. The sensing system of claim 1 , wherein the measurement model to be taught is in the form of mobile code.
4. The sensing system of claim 1 , wherein the first phenomenon state includes a state of one or more of: a sound, motion, vibration, altitude, luminous intensity, proximity, pressure, temperature, radiation, timing, humidity, electromagnetic field intensity, altitude, weight, airborne particulates, chemical composition, velocity, direction, and distance.
5. A method, comprising:
sampling, at a respective sensor in a set of sensors, a measurement value, wherein the sensor stores a first measurement model as P(Z=z|X=x) for generating a likelihood function for a first phenomenon state, wherein Z corresponds to the measurement value and X corresponds to a first phenomenon state under investigation;
generating likelihood functions at the sensor by placing the sampled measurement value into the first measurement model;
collecting the likelihood functions from the set of sensors; and
computing an aggregate probability of a second phenomenon state based on
∏
i
=
1
n
P
(
Z
i
=
z
i
❘
X
=
x
)
which corresponds to a product of a number of n likelihood collected from the sensors.
6. The method of claim 5 , wherein collecting the likelihood functions comprises communicating the likelihood functions in the form of mobile code.
7. The method of claim 5 , wherein the measurement model to be loaded and/or reloaded is in the form of mobile code.
8. The method of claim 5 , wherein the first phenomenon state includes a state of one or more of: sound, motion, vibration, altitude, luminous intensity, proximity, pressure, temperature, radiation, timing, humidity, electromagnetic field intensity, altitude, weight, airborne particulates, chemical composition, velocity, direction, and distance.
9. A non-transitory computer-readable storage medium including instructions that when executed by a computer system cause the computer to perform a method, the method comprising:
sampling, at a respective sensor in a set of sensors, a measurement value, wherein the sensor stores a first measurement model as P(Z=z|X=x) for generating a likelihood function for a first phenomenon state, wherein Z corresponds to the measurement value and X corresponds to a first phenomenon state under investigation;
generating likelihood functions at the sensor by placing the sampled measurement value into the first measurement model;
collecting the likelihood functions from the set of sensors; and
computing an aggregate probability of a second phenomenon state based on
∏
i
=
1
n
P
(
Z
i
=
z
i
❘
X
=
x
)
which corresponds to a product of a number of n likelihood collected from the sensors.
10. The non-transitory computer-readable storage medium of claim 9 , wherein collecting the likelihood functions comprises communicating the likelihood functions in the form of mobile code.
11. The non-transitory computer-readable storage medium of claim 9 , wherein the measurement model to be loaded and/or reloaded is in the form of mobile code.
12. The non-transitory computer-readable storage medium of claim 9 , wherein the first phenomenon state includes a state of one or more of: sound, motion, vibration, altitude, luminous intensity, proximity, pressure, temperature, radiation, timing, humidity, electromagnetic field intensity, altitude, weight, airborne particulates, chemical composition, velocity, direction, and distance.
13. The system of claim 1 , further comprising a teaching mechanism configured to load a second measurement model onto the sensor, wherein the second measurement model configures the sensor to determine a likelihood function for a third phenomenon state based in part on the sampled measurement value.
14. The sensing system of claim 13 , wherein the teaching mechanism resides with the data-fusing mechanism.
15. The method of claim 5 , further comprising:
loading a second measurement model onto the sensor, wherein the second measurement model configures the sensor to determine a likelihood function for a third phenomenon state based in part on the sampled measurement value.
16. The non-transitory computer-readable storage medium of claim 9 , wherein the method further comprises:
loading a second measurement model onto the sensor, wherein the second measurement model configures the sensor to determine a likelihood function for a third phenomenon state based in part on the sampled measurement value.