Research On Underground Localization Based On Online Dynamic Correction Of Radio Map

Since underground personnel needs to be located accurately to ensure coal mine security, underground positioning has been a hot topic in recent years. This paper studies the received signal strength (RSS)-based positioning technique according to the features of the underground environment, and focuses on two problems from engineering and theoretical perspective.First it demonstrates the engineering feasibility of constructing an underground positioning network and proposes a WLAN-based underground localization system. We investigate the technological advantages and feasibility of position fixing through the 2.4G signals and the ZigBee network. Also we design the circuit and controlling software of the transmitters for blind nodes in the underground positioning network, and test the transmitter.Second, it proposes a radio map dynamic calibration based underground location algorithm (RM-DA-UL). It starts from estimating the RSS observations with the Bayes regression algorithm. We use the estimated values rather than the measured values to construct the radio map. In this way, the influence of noise and multipath interferences on position fixing is eliminated effectively. In the radio map online dynamic correction stage, it provides accurate standard deviations of the locations. In the positioning stage, it assigns high weights for most of the locating points in proximity to the radio map using the weighted KNN algorithm. Then, it computes the average of weights to select an appropriate AP anchor and locate the blind node.Experimental results show that the proposed blind node transmitter is capable of underground positioning at an expected accuracy. The proposed system greatly outperforms existing state-of-the-art algorithms in terms of accuracy, failure rate, number of localized objects and the anti-interference ability.