Undergraduate_Thesis

My repo for my undergraduate thesis at THU. I propose a self-localization strategy for sensor networks based on MDS initialization，Adam optimization and ICP aligning. It works well for both static and dynamic networks in simulations.

Static network localization

2D space (10mx10m map):

3D space (10mx10mx10m map):

Algorithm performance:

Very large scale network (100~1000 nodes)

In 100mx100m map:

---

Dynamic network localization

Without noise:

With noise:

Add a naive Kalman Filter: