Motion detection in mesh networks
In a general aspect of the examples described, motion is detected based on received wireless signals in a mesh network. In an example, transfer function components for each pair of devices are obtained. A time series of transfer function matrices is generated, with each of the transfer function matrices comprising a respective subset of the transfer function components. Motion of an object in the space is detected based on the time series of transfer function matrices.
1. A method of motion detection in a mesh network, comprising:
obtaining transfer function components for each pair of devices in the mesh network, the transfer function components for each pair of devices based on wireless signals communicated through a space between the pair of devices;
generating a time series of transfer function matrices, each transfer function matrix comprising a respective subset of the transfer function components; and
detecting, by operation of one or more processors in a motion detection system, motion of an object in the space based on the time series of transfer function matrices.
2. The method of claim 1 , wherein each subset of transfer function components comprises the transfer function components obtained for a certain transfer function matrix in the time series.
3. The method of claim 1 , wherein each matrix position in each of the transfer function matrices represents a wireless link between one of the respective pairs of devices.
4. The method of claim 1 , wherein each of the transfer function matrices is an M×M matrix, and the mesh network comprises M wireless devices.
5. The method of claim 1 , wherein detecting motion comprises processing the time series of transfer function matrices by operation of a neural network.
6. The method of claim 1 , wherein detecting motion comprises detecting differences of transfer function components at one or more matrix positions over the time series.
7. The method of claim 6 , wherein detecting differences of transfer function components comprises determining differences between transfer function components at adjacent points of the transfer function matrices in the time series.
8. A motion detection system comprising:
one or more processors;
memory storing instructions, which when executed by the one or more processors, cause the one or more processors to:
obtain transfer function components for each pair of devices in a mesh network, the transfer function components for each pair of devices based on wireless signals communicated through a space between the pair of devices;
generate a time series of transfer function matrices, each transfer function matrix comprising a respective subset of the transfer function components; and
detect motion of an object in the space based on the time series of transfer function matrices.
9. The motion detection system of claim 8 , wherein each subset of transfer function components comprises the transfer function components obtained for a certain transfer function matrix in the time series.
10. The motion detection system of claim 8 , wherein each matrix position in each of the transfer function matrices represents a wireless link between one of the respective pairs of devices.
11. The motion detection system of claim 8 , wherein each of the transfer function matrices is an M×M matrix, and the mesh network comprises M wireless devices.
12. The motion detection system of claim 8 , wherein detecting motion comprises processing the time series of transfer function matrices by operation of a neural network.
13. The motion detection system of claim 8 , wherein detecting motion comprises detecting differences of transfer function components at one or more matrix positions over the time series.
14. The motion detection system of claim 13 , wherein detecting differences of transfer function components comprises determining differences between transfer function components at adjacent points of the transfer function matrices in the time series.
15. A non-transitory computer readable medium comprising instructions which when executed by one or more processors cause the one or more processors to:
obtain transfer function components for each pair of devices in a mesh network, the transfer function components for each pair of devices based on wireless signals communicated through a space between the pair of devices;
generate a time series of transfer function matrices, each transfer function matrix comprising a respective subset of the transfer function components; and
detect motion of an object in the space based on the time series of transfer function matrices.
16. The non-transitory computer readable medium of claim 15 , wherein each subset of transfer function components comprises the transfer function components obtained for a certain transfer function matrix in the time series.
17. The non-transitory computer readable medium of claim 15 , wherein each matrix position in each of the transfer function matrices represents a wireless link between one of the respective pairs of devices.
18. The non-transitory computer readable medium of claim 15 , wherein each of the transfer function matrices is an M×M matrix, and the mesh network comprises M wireless devices.
19. The non-transitory computer readable medium of claim 15 , wherein detecting motion comprises processing the time series of transfer function matrices by operation of a neural network.
20. The non-transitory computer readable medium of claim 15 , wherein detecting motion comprises detecting differences of transfer function components at one or more matrix positions over the time series.
21. The non-transitory computer readable medium of claim 20 , wherein detecting differences of transfer function components comprises determining differences between transfer function components at adjacent points of the transfer function matrices in the time series.