英文摘要 | With the development of positioning technology, more and more people are experiencing the convenience of location-based services. Wireless sensor networks can easily get the location information, and help to improve the intelligence of the applications. In this paper, we studied localization in wireless sensor networks, the study addressed to the hardware, software as well as positioning algorithms. We designed a small size portable node, and implemented the network between nodes. Then we analyzed the radio signal characteristics, and established a radio channel model. Finally, we used the radio channel model in an improved RSS fingerprint search algorithm, and designed a distributed weighted centroid algorithm. Our research contains the following categories: 1. Designed the node's electric circuit, including the processor part and the RF communication part. We added some new features to the node, such as low power alarm. When the battery voltage is too low, the low power alarm circuit can notify the processor its battery state. We also multiplexed using the charging interface and the programming interface using a reed switch, so that the quantity of plugs can be reduced. Finally, we added a button to the node. The node's size is 48*40.5*4mm, only a little thicker than a one yuan coin. 2. Designed a fixed routing protocol and mobile node positioning program framework. We designed a fixed-route communication protocol, using user defined routing table to route a package between the nodes. We also designed two different positioning frameworks, one is the centralized positioning frame, in which the node will send all the information to the server, and the server compute the location of the mobile node; the other is distributed positioning framework, the node directly calculate the position in its own processor. 3. We studied radio channel characteristics of wireless sensor networks, analyzed a variety of different factors impact on the wireless signal. Then we designed an empirical mode decomposition (EMD) modeling method. This method only use a small amount of data in the experimental environment, uses a data-driven modeling, and calculates the model that adapts to the environment. Furthermore, we analyzed the static characteristics of the model. 4. Based on the RSS model, we designed two positioning methods. One is centralized computing method, using EMD model in the improved RSS fingerprint search method. Our experiment shows that the improved method is... |
修改评论