Suppose there is a rectangle indoor area, we want to locate different positions within this area using TDoA estimations. 5 sensors are placed to obtain optimal 3D positions with TDoA errors, we only care about horizontal positions (i.e., x and y coordinates), while the error in z coordinate estimation is the least thing we care. We place the first sensor at the origin and implement linear least square algorithm to get the optimal estimated position.
The above is the problem background, my question is: what's the optimal geometric node placement under this condition? By saying "optimal" I am talking about the optimal placement that maximises the TDoA error tolerance ability. So is there an optimal node placement pattern that can maximise the TDoA error tolerance? I have seen many papers regarding to the 2D case, but so far I didn't see one for 3D positioning.