CODEWITHSUNDEEP
locationclassificationbluetooth-lowenergyrssi
st20000428 Harith
st20000428 Harith

Reputation: 47

Improve location using Bluetooth?

For my project i need estimate the point on a grid that i use. To check my method works i took some Readings through x axis as below when y=2;enter image description here

Blue, Brown and Grey are the access point 1,2 and 3 average RSSI readings. The node moves from red dot to blue which is 120 cm.The fluctuations of the RSSI readings are not linear and this is a very big problem in my case to get the accurate position. I use Knn to get the nearest position. What can i do to make it correct.? Use some other classifier will help ?

Upvotes: 0

Views: 79

Answers (1)

dge
dge

Reputation: 3075

Check out Is rssi a reliable parameter in sensor localization algorithms: An experimental study

While i dont completely agree to the way those tests were executed and analyzed, they miss alot of details on analyzing the results such as differentiating the readings of RSSI over the different BLE channels, or measurements on antenna characteristics and orientation, the core statement i consider quite on point.

RSSI cannot be used as a reliable metric in localization algorithms

Beside the issues reflection, shadowing, antenna characteristics, etc. pose on the difficulty, the BLE Specification itself adds to the problem as the RSSI is not defined as an absolute value, it is specified to be used in a relative manner related to the distance to the golden receiver range, the rx power that would be not too weak and not too strong for the receiver to have the best receive quality. Also this RSSI can vary +/- 6dBm from the real value.

This means, we can hardly rely on the same readings across different devices and secondly, the accurracy is allowed to vary alot according to the specification.

For that reason, projects relying too hard on that accurracy are doomed to fail one way or another. However there are still applications possible getting something positive out of these RSSI readings, i.e. not relying completely on them, but instead use them as indicator in a supportive way.

If you are interested more on this matters, search for Indoor localization rssi i.e. on google scholar.

Upvotes: 1

Related Questions