Estimating qualities of links between nodes of a wireless network

1. A method of estimating qualities of links between nodes, comprising:

passively receiving, by a plurality of network sensors configured to determine link quality, a plurality of wireless packets between a network device and an access point to the wireless network;

obtaining, by a service engine, a plurality of link quality signatures from the plurality of network sensors, wherein each of the plurality of link quality signatures includes capacity and strength link qualities between a network node, the network node including the network device or the access point, and the plurality of network sensors, wherein the obtained plurality of link quality signatures are determined upon sensing a motion of the network node of greater than a threshold;

estimating, by the service engine, the capacity and strength link qualities between the network device and the access point based on training measurements on the plurality of link quality signatures and an estimation model;

determining the estimation model by assigning different weights to different cases of the training measurement, wherein assigning the different weights includes assigning additional weight to the different cases of the training measurement if the capacity and strength link qualities are below a threshold; and

wherein obtaining the plurality of link quality signatures by the service engine is based on a sampling rate of the plurality of wireless packets, and wherein a plurality of coordinated network sensors of the plurality of network sensors is synchronized relative to the sampling rate to improve data throughput to the service engine.

2. The method of claim 1 , further comprising obtaining, by the service engine a stream of pluralities of link quality signatures.

3. The method of claim 2 , further averaging multiple of the stream of pluralities of link quality signatures.

4. The method of claim 1 , wherein each signature of the plurality of link quality signatures is based on a wireless packet of the plurality of wireless packets passively received by the plurality of network sensors.

5. The method of claim 1 , wherein obtaining the plurality of link quality signatures comprises:

receiving a wireless packet descriptor from each of the plurality of network sensors;

comparing the wireless packet descriptors to determine which wireless packets were commonly received from the network node by the plurality of network sensors.

6. The method of claim 5 , wherein the wireless packet descriptor includes a packet transmitter address.

7. The method of claim 5 , wherein the wireless packet descriptor includes a timestamp of when the wireless packet was received by a network sensor of the plurality of network sensors.

8. The method of claim 1 , further comprising determining the estimation model based on wireless propagation equations.

9. The method of claim 1 , further comprising determining the estimation model comprising characterizing wireless propagation based on training measurements of link quality signatures from a second plurality of network sensors and corresponding training measurements of link quality from a second plurality of network nodes.

10. The method of claim 1 , further comprising determining the estimation model comprising characterizing wireless propagation with physical equations and characterizing wireless propagation based on prior measurements of link quality signatures and corresponding link quality between a second plurality of network nodes.

11. The method of claim 1 , wherein the estimation model is precomputed for each of a plurality of different scenarios, and one of the estimation models is selected for each deployment.

12. The method of claim 1 , further comprising determining the estimation model comprising:

modeling network sensor operation after deployment of the plurality of network sensors.

13. The method of claim 12 , wherein the estimation model is initialized before deployment, and adapted to a deployment site based on a set of measurements performed by the plurality of network sensors.

14. The method of claim 12 , wherein the estimation model continually adapts to additional measurements that become available while deployed.

15. The method of claim 1 , wherein estimating the link qualities between the plurality of network nodes is further based on link quality signatures between the plurality of network sensors.

16. The method of claim 15 , wherein the link quality signatures between the plurality of network sensors is further based on a subset of the plurality of network sensors transmitting wireless packets and other network sensors of the plurality of network sensors receiving and estimating a link quality between the plurality of network sensors.

17. The method of claim 16 , wherein the link quality signatures between the network sensors is determined periodically.

18. The method of claim 1 , further comprising preprocessing each of the plurality of link quality signatures before estimating the link qualities, and estimating the link qualities between the plurality of network nodes based on the plurality of preprocessing link quality signatures and the estimation model.

19. The method of claim 18 , wherein preprocessing each of the link qualities of each signature of each network sensor comprises determining a ratio of the link quality between the network sensor and the network node and a link quality between a selected network sensor and the network node, thereby accounting for variations in transmit power of the network node over time.

20. A wireless network comprising:

a plurality of network nodes;

a plurality of network sensors configured to determine link quality, wherein the plurality of network sensors passively receive a plurality of wireless packets between a network device and an access point to the wireless network;

a service engine, wherein the service engine is operative to:

obtain a plurality of link quality signatures from the plurality of network sensors, wherein each signature includes capacity and strength link qualities between a network node, the network node including the network device or the access point, and the plurality of network sensors, wherein the plurality of link quality signatures is determined upon a motion of the network node of greater than a threshold;

estimate the capacity and strength link qualities between the network device and the access point based on training measurements on the plurality of link quality signatures and an estimation model;

determine the estimation model, wherein determining the estimation model includes assigning different weights to different cases of the training measurements, wherein assigning the different weights includes assigning additional weight to the different cases of the training measurement if the capacity and strength link qualities are below a threshold,

rebalancing the training measurements using the different weights; and

wherein obtaining the plurality of link quality signatures by the service engine is based on a sampling rate of the plurality of wireless packets, and wherein a plurality of coordinated network sensors of the plurality of network sensors is synchronized relative to the sampling rate to improve data throughput to the service engine.