Research On Outdoor Passive Localization In WSNs

Wireless sensor networks are widely used in various applications, such as military affairs, aviation, environment monitoring, medical care, residential sensing, industrial and commercial areas. In real word applications, the measurements can be valuable and meaningful only when the event is correlated with the location information. Thus, it’s practically significant and valuable to study the localization technology in WSNs.Traditional localization technology (i.e., active localization) needs the target to be attached with device (e.g., GPS models and RFID tags). However, this kind of technology is not available in some scenario, such as wild animal protection and intrusion detection, since targets can’t carry a device. Passive localization has been widely studied and applied since it does not require target carry any device. Besides, it does not require the line of sight and the visual angle, which makes the data procedure simpler.Recent years have witnessed the emergence of many remarkable researches in passive localization. However, when we apply it to complex outdoor scenarios, there are still some challenges:(1) The localization would be failed due to the loss and imperfectness of RSSI data in complex outdoor scenarios; (2) The RSSI value fluctuates in complex outdoor scenarios, and such characteristic would lead to the mismatch between the online measurement and the prior knowledge, which results in high localization error; (3) Current approaches performance poorly in some scenes that the prior knowledge is unavailable; (4) In a heterogeneous data sources scene, how to improve the localization accuracy with different characteristics of the data sources. To cope with the above problems, this paper provides a passive localization method which is especially suitable for outdoor scenarios, in addition we provides-theoretical analysis and support for a more efficient localization approach.The most innovative researches in this paper:(1) Proposes a passive localization approach based on double level fusion modelTo cope with the loss and imperfectness of RSSI data in the outdoor complex environments, we studies the spatial-temporal feature of RSSI, and then designs a double level fusion model using the Bayesian network. To solve the problem that targets can’t be detected in strong noise and unreliability channel scene, this paper designs a sliding average based detection algorithm in data fusion level by taking the advantage of partial stability of RSSI. The decision fusion level improves the accuracy by using the Bayesian network to make up the temporal diversity of RSSI. The real world experimental results demonstrate that this approach is feasible and we also analyze the localization accuracy.(2) Proposes a passive localization approach based on probability estimation modelThe RSSI value fluctuates in complex outdoor scenarios, and such characteristic would lead to the mismatch between the online measurement and the prior knowledge, which results in high localization error. To solve this problem, this paper proposes a "two-phase" localization approach. It first localizes the target using an independent space estimation. Then, it exploits the mean value of the space and the time to eliminate error locations and improve the accuracy by interpolation for estimated locations. We evaluate its validity and effectiveness through extensive experiments(3) Propose a passive localization approach based on signal strength using signal propagation modelTo cope with the problem that current approaches performance poorly in some scenes that the prior knowledge is unavailable, this paper analyses the wireless signal propagation model, scattering model and diffraction model, and proposes a self-adaptive link performance aware localization method. This paper discusses the localizability and designs the unit localization area to assure that target can be localized uniquely. Compared with other methods, our method can largely enlarge the localization area without increasing the data type. Extensive simulation results illustrate the superior performance of the proposed method in terms of the localizability and the accuracy.(4) Proposes a passive localization approach based on heterogeneous data synergistic methodTo solve the problem that there are heterogeneous data sources in the network, this paper combines signal strength and infrared ranging method and utilizes their different characteristics to localize targets. According to the Bayesian network and the different characteristics, this paper proposes a localization model which is suitable for outdoor environment. The localization accuracy is improved by using a particle filter method. Simulation results prove that the validity of our passive localization method-and prove that the localization accuracy is greatly improved compared with other methods.