Localization of a Mobile Device with Sensor Using a Cascade Artificial Neural Network-Based Fingerprint Algorithm
International Journal of Computational Intelligence Systems
One of the important functions of sensor networks is that they collect data from the physical environment and transmit them to a center for processing. The location from which the collected data is obtained is crucial in many applications, such as search and rescue, disaster relief, and target tracking. In this respect, determination of the location with low-cost, scalable, and efficient algorithms is required. This study presents the implementation of a fingerprint-based location determination
... algorithm by using the cascade artificial neural network (ANN). A 15.6 × 13.8 m 2 implementation area, in which an anchor node is placed at each corner, is divided into grids with a 60-cm edge. The proposed algorithm consists of two phases: offline and online. In the offline phase, first a mobile device with an Xbee sensor, which is able to move sensitively and communicate with anchor nodes, is used. With this device, the implementation area is visited, and at each grid point, received signal strength indicator (RSSI) values and real distances measured from the anchor nodes are recorded in a database. The training of the cascade ANN is done using the database for both range and location determination. In the online phase, the RSSIs measured by the anchor nodes are provided as the input to the cascade ANN algorithm by means of a mobile device in any coordinate. The location of the mobile device and its distance to the anchor nodes are determined with minimum error. To show the superiority of the proposed method, the results obtained are compared with those in the literature and it has been shown that this location determination is made with a smaller error. the location determination is important problems for WSNs. Global positioning systems (GPSs) are used for location determination  . However, GPSs are not ideal for many applications because they work only in open areas, they are costly, and consume high power. To eliminate this disadvantage, a determination of the location in closed environments can be calculated by means of the three sensor nodes, whose locations are known. In general, localization accuracy can be improved by increasing the number of anchor nodes. However, the cost of an anchor node is higher than that of ordinary nodes. Therefore, it is necessary to reduce the number of anchor nodes involved in node localization. Reducing the number of anchor nodes decreases the localization cost, but increases the location error and decreases accuracy. ID:p0080 Localization determination is performed in two steps. The first step is to estimate the distance between sensors. In this step, the distance is estimated by means of the characteristics of the signal used for the communication between sensors. The second step is the location determination. This step is performed by using various methods, such as centroid localization (CL), weighted centroid localization (WCL), and the bounding box method. ID:p0085 Localization algorithms in close environments are basically divided into two groups: range-based and range-free methods [6, 7] . Figure 1 shows the classification of localization algorithms. ID:p0095 The main parameter used by range-based localization methods in location determination is distance.