Hi,
let me clarify some points about localization and ns-3, as it seems that there's something that is not understood at all.
Nodes in ns-3 knows their exact position and the exact time.
Localization is all about having an imperfect knowledge of position and time.
You understand now what is the problem ?
Time: we recently added some functions to have a per-node clock, each one with its own precision. The functionality is still at extremely early stage, and right now the API is just a wrapper around the "old" absolute time. As a consequence, one should first fill this gap. Any contribution is welcome.
Position: as for time, node position is down by nodes exactly. However, one should add a imperfect knowledge layer, e.g., one taking into account the relative error of GPS, or something similar.
MOREOVER, localization techniques are affected by non-deterministic delays in signal processing. I.e., when you receive a signal a node could have an internal random processing delay on top of the fixed and known part. This adds to the localization precision outcomes. If you don't add this too, your results will be flawed. Note that processing delay is not modeled at all in ns-3.
Summarizing: good luck. It is possible to simulate localization using ns-3, but you should have a deep understanding of the whole problem and of the limits / restrictions of the simulation environment. Fail at one small bit and your results are invalid.
Have fun.
T.
PS: the above considerations apply to Matlab too, and any other simulation environment.