Cooperative Localization on Computationally Constrained Devices.

摘要

Cooperative localization is a useful way for nodes within a network to share location information in order to better arrive at a position estimate. This is handy in GPS contested environments (indoors and urban settings). Most systems exploring cooperative localization rely on special hardware, or extra devices to store the database or do the computations. Research also deals with specific localization techniques such as using Wi-Fi, ultra-wideband signals, or accelerometers independently opposed to fusing multiple sources together. This research brings cooperative localization to the smartphone platform, to take advantage of the multiple sensors that are available. The system is run on Android powered devices, including the wireless hotspot. In order to determine the merit of each sensor, analysis was completed to determine successes and failures. The accelerometer, compass, and received signal strength capability were examined to determine their usefulness in cooperative localization. Experiments at meter intervals show the system detected changes in location at each interval with an average standard deviation of 0.44m. The closest location estimates occurred at 3m, 4m and 6m with average errors of 0.15m, 0.11m, and 0.07m respectively. This indicates that very precise estimates can be achieved with an Android hotspot and mobile nodes.