0

Today is one of those days when the correct method to calculate this escapes me. I have 3 fixed points on a map where measurement devices are located. These all track the same transmitters. A mobile unit operates somewhere between the measurement devices. This too tracks the transmitter. I have a modeled delay calculation for each of the fixed receivers. I want to form a delay value for the reported position of the mobile unit. I have delay values, exact coordinates, distance to the mobile unit at the time of last mobile reported position. Just trying to decide how to apply the values in some kind of weighted system to get an accurate value for the mobile position.

J Hinton
  • 1

1 Answers1

0

A kalman filter would definitely do the trick. Each of the measurements could be utilized in a sensor fusion method. Using a kalman filter for the data as a preprocess to the system state update is an option.

A number of implementations are published.

Daniel Fischer
  • 206,697
Derek Kuether
  • 11
  • I can see how that would work if I was tracking previous results. I was hoping to perform this as a stand alone single caculation so I didn't need to maintain more variables in software. The delay for the fixed locations is a smoothed running calculation so I'm already using a bit of RAM. – J Hinton Jun 22 '15 at 12:06
  • Do expect the weights to remain constant? – Derek Kuether Jun 22 '15 at 13:00
  • Also is the measurement an exact position or do you need to use the other measurement sensors to triangulate a position?

    Kalman filter can be used for a single step merging of data. Let's assume you do not need to triangulate. Suppose you have measurement one and measurement two with accompanying variances. Let measurement one be your initial state estimation.

    M_1, M_2, V_1, V_2,

P_i = V_1, X_i = h(M_1).

Use the next measurement as the "measurement" to merge variances 
and measurements.  See the link in the answer for detailed   
explanation on how to do this.
    – Derek Kuether Jun 22 '15 at 14:10
  • The mobile unit would report its current position every few seconds. The fixed units have exact coordinates. The weights would change as the mobile unit moves and distances between the mobile and fixed sensors changes. The fixed sensor values will change but with a gradual shift. Its been many many years since I had any upper level math so grasping the kalman filter could take me some time. I've looked over the linked page and am trying to figure out the implementation that I could use here. – J Hinton Jun 22 '15 at 17:58