System for detection of body motion
Methods for classifying aggregated data in a distributed sensor system are provided and illustrated with a wearable motion sensor network. The classification is operated in a distributed fashion on individual sensor nodes and a base station computer. The method classifies actions using a set of training motion sequences as prior examples and may reject outlying actions that are not in the training categories. Acquired sensor data is processed at the node by taking projections of the data to reduce dimensionality, calculating sparse representations of features using training sequences; validating and classifying local measurements and then transmitting classified measurements to a network base station. The base station aggregates local sensor measurements and performs a global classification of the data by forming global features from the local measurements; calculating sparse representations of global features; validating and classifying valid global features; and labeling global features and their corresponding local features.
1. A method for classifying aggregated data in a distributed sensor system, comprising:
obtaining sensor data from a plurality of sensors at a local site;
performing a local classification of the sensor data using local processing, the local processing comprising:
taking projections of the sensor data to reduce dimensionality;
calculating sparse representations of features using training sequences;
validating local measurements;
classifying valid local measurements; and
transmitting classified valid measurements to a network base station.
2. The method as recited in claim 1 , wherein a projection {Ã} j =R j A j and {{tilde over (y)}} j =R j Y j is taken for each sensor j to reduce dimensionality of the obtained training sensor data and query measurements.
3. The method as recited in claim 1 , wherein the sparse representation of measurement feature {{tilde over (y)}} i based on training dictionary {Ã} j is solved with x*=argmin i ∥x∥ 1 subject to {tilde over (y)} j =Ã j x.
4. The method as recited in claim 1 , wherein the validation of local measurements comprises:
comparing local measurements to a sparsity concentration index (SCI); and
validating local measurements that are above an SCI threshold.
5. The method as recited in claim 4 , wherein the sparsity concentration index (SCI) is defined by
SCI
(
x
)
=
.
K
·
max
j
=
1
,
K
,
K
δ
j
(
x
)
1
/
x
1
-
1
K
-
1
∈
[
0,1
]
,
where K is the number of classes.
6. The method as recited in claim 1 , further comprising:
transmitting labeled valid local measurements from a local processor to a base station receiver.
7. The method as recited in claim 1 , further comprising:
aggregating valid local sensor measurements from one or more sensors at a network base station; and
performing a global classification of the aggregated local sensor data, the global classification comprising:
forming global features from valid local measurements from a plurality of sensors;
calculating sparse representations of global features;
validating global features;
classifying valid global features; and
labeling global features and their corresponding local features.
8. The method as recited in claim 7 , wherein the sparse representation of a global feature is solved based on a global dictionary {Ã}′ with the equation: x*=arg min∥x∥ 1 subject to {tilde over (y)}′=A′x.
9. The method as recited in claim 7 , wherein the validation of global features comprises:
comparing global measurements to a sparsity concentration index (SCI); and
validating global measurements that are above an SCI threshold.
10. The method as recited in claim 7 , wherein the label is recovered according to the equation: label ({y}′)=argmin i=1, . . . ,K ∥{{tilde over (y)}}′−Ã′δ i (x)∥.
11. A method for classifying aggregated data in a distributed sensor system, comprising:
obtaining sensor data from a plurality of sensors at a local site;
performing a local classification of the sensor data with local processing, the local processing comprising:
taking projections of the obtained sensor data to reduce dimensionality;
calculating sparse representations of measurements using training sequences;
validating local measurements; and
labeling valid local measurements;
transmitting labeled valid local measurements from a local processor to a base station receiver; and
performing a global classification of the labeled sensor data using global processing with a base station processor, the global processing comprising:
forming global features from valid local measurements from a plurality of sensors;
calculating sparse representations of global features;
validating global features;
classifying valid global features; and
labeling valid global features and corresponding local features.
12. A distributed sensor system, comprising:
a plurality of sensor nodes, each node having a node processor, a plurality of sensors and a transceiver;
a non-transitory processor-readable medium including one or more instructions for:
obtaining sensor data from a plurality of sensors at a local site;
performing a local classification of the sensor data with local processing on the node processor, the local processing comprising:
taking projections of the sensor data to reduce dimensionality;
providing training sequences for each sensor;
calculating sparse representations of features using training sequences;
validating local measurements; and
classifying valid local measurements;
a base station with a station processor and a transceiver; and
a non-transitory processor-readable medium including one or more instructions for:
obtaining labeled sensor data transmitted from a plurality of sensor nodes;
performing a global classification of the sensor data with the station processor, the global processing comprising:
forming global features from valid local measurements from a plurality of sensors;
calculating sparse representations of global features;
validating global features; and
classifying valid global features; and
labeling valid global features and corresponding local features.