The wireless localization techniques disclosed herein may be combined with other sensed information to improve localization accuracy of the overall network. For example, in wireless sensors in which one or more of the nodes contain cameras, captured images can be used with image processing and machine learning techniques to determine whether the sensor nodes that are being monitored are looking at the same scene and are therefore likely in the same room. Similar benefits can be achieved by using periodic illumination and photodetectors. By strobing the illumination and detecting using the photodetectors, the presence of an optical path can be detected, likely indicating the absence of opaque walls between the strobe and the detector. In other embodiments, magnetic sensors can be integrated into the sensor nodes and used as a compass to detect the orientation of the sensor node that is being monitored. This information can then be used along with localization information to determine whether the sensor is on the wall, floor, ceiling, or other location.

In one example, each sensor node may include an image sensor and each perimeter wall of a house includes one or more sensor nodes. A hub analyzes sensor data including image data and optionally orientation data along with localization information to determine absolute locations for each sensor node. The hub can then build a three dimensional image of each room of a building for a user. A floor plan can be generated with locations for walls, windows, doors, etc. Image sensors may capture images indicating a change in reflections that can indicate home integrity issues (e.g., water, leaking roof, etc.).