A localization strategy based on n-times trilateral centroid with weight

Tie Qiu, Yu Zhou, Feng Xia, Naigao Jin, Lin Feng
2012 International Journal of Communication Systems  
Localization based on received signal strength indication (RSSI) is a low cost and low complexity technology, and it is widely applied in distance-based localization of wireless sensor networks. The error of existing localization technologies is significant. This paper presents the N-times trilateral centroid weighted localization algorithm, which can reduce the error considerably. Considering the instability of RSSI, we use the weighted average of many RSSIs as current RSSI. To improve the
more » ... racy, we select a number of (no less than three) reliable beacon nodes to increase the localization times. Then we calculate the distances between reliable beacon nodes and the mobile node using an empirical formula. The mobile node is located N times using the trilateral centroid algorithm. Finally, we take the weighted average of the filtered reference coordinates as the mobile node's coordinates. We conduct experiments with the STM32W108 chip, which supports IEEE 802.15.4. The results show that the proposed algorithm performs better than the trilateral centroid algorithm. LOCALIZATION BASED ON N -TIMES TRILATERAL CENTROID WITH WEIGHT 1161 of arrival [5] measures the distances between nodes by calculating the transmission time of one kind of signal. Time difference of arrival [6] measures the distances by calculating the arrival-time interval of two different signals. Angle of arrival [7] measures the arrival-direction of signals with antenna array, and then calculates the relative orientation or angle according to the measured arrivaldirection. Finally, it uses the triangulation method to calculate the mobile node's position. These three methods require additional hardware support, which increases the spending. The methods in [8] [9] [10] are all distance-independent localization algorithms. Centroid algorithm [8] uses the geometric center of a mass of beacon nodes as the unknown nodes' locations. This method is simple but the error is large, so it is impractical. The DV-Hop algorithm [9] uses hop counts to denote distance and the error is great. The unknown node calculates the distances with beacon nodes by measuring the hop counts with beacon nodes, and then uses the measured distance to locate itself. The approximate point-in-triangulation test algorithm [10] locates the unknown node via continuously identifying whether the unknown node is within the triangle area, which is structured by three beacon nodes. This method can only calculate the position roughly. The error of the mentioned localization methods is big so the methods are not suitable for accurate localization. The methods in [11] [12] [13] are all put forward based on trilateration. Localization accuracy can be improved if we take weight into account for trilateral localization algorithm [11] or use a modified model to correct the measured RSSI [12, 13] . Using many groups of historical RSSI to calculate current RSSI can reduce the influence of instability of RSSI [14] , but it does not consider the weight of historical value. We improve it by taking the weight of historical value into account to calculate current RSSI. The authors in [15] use the sum of the measured distances' reciprocal instead of the reciprocals of the measured distances' sum as the weight and puts forward correction-factor to avoid the information inundation phenomenon. This new method of calculating weight ensures that with the increase of the distances between nodes, the weight is decreased. The received signal strength difference algorithm [16] defines different weights for different beacon nodes based on RSSI. Then it uses the weighted centroid algorithm to achieve location. Beacon nodes' layout has an important influence on the localization accuracy and the best layout of beacon nodes is an equilateral triangle. The authors in [17] improved the centroid algorithm to ellipse centroid localization algorithm and at the same time combined it with error factor and precision factor. The reliability of the beacon node is high if it is close to the mobile node and the accuracy of localization is high if localization is performed with a reliable beacon node. Triangle or polygon centroid localization algorithm selects the three or more beacon nodes that are closest to the mobile node to locate the mobile node [18] . The localization accuracy of taking the link quality indicator value into account during the weighted centroid localization is better than the weighted centroid localization algorithm [19] . The method mentioned in [20] divides the block into different blocks according to the relative positions of beacon nodes and mobile node, and then defines different priorities for different blocks. Different priorities represent different weights. This is a new way to determine the weight for beacon nodes. However, the accuracy of this method is significantly influenced by the measured accuracy of RSSI. The polygon algorithm [21] is better than trilateration. The way to get better parameters for the empirical formula that converts RSSI to distance is described in detail in [21] and we propose a new method to get parameters better than in [21] . The authors in [22] dealt with wireless sensor networks based on IEEE802.15.4 protocol and analyzed three indoor environment localization methods based on RSSI: trilateration, minimax algorithm, and maximum likelihood estimation. The experiment results show that in the indoor environment the trilateration shows excellent properties and the availability is good. There are many efficient algorithms but many of them require additional hardware support. Localization methods based on RSSI do not need any additional hardware support and do not have to keep synchronous with the network. Compared with distance-independent algorithms, distance-dependent algorithms are more accurate. The trilateral centroid algorithm uses RSSI to calculate the distances between nodes and it is a distance-dependent algorithm. Also, it is better than trilateration and centroid positioning method. To improve the localization accuracy, we propose the N -times trilateral centroid weighted localization algorithm (NTCWLA, N denotes the times of using trilateral centroid algorithm) [23] . We select n.n > 3/ reliable beacon nodes and then combine any three of them to calculate the reference coordinates of the mobile node with trilateral Tie Qiu is a lecturer and PhD candidate in computer science at the Dalian University of Technology, China. His research interests cover embedded system architecture, wireless sensor networks, and systems modeling. He is a member of the China Computer Federation and ACM. Yu Zhou received his BE degree from the Dalian University of Technology, China. Currently, he is a master's student in the School of Software, Dalian University of Technology. His research interests cover wireless sensor networks and Internet of Things. Feng Xia is an associate professor and a PhD supervisor in the School of Software, Dalian University of Technology, China. He is the (Guest) Editor of several international journals. He serves as General Chair, PC Chair, Workshop Chair, Publicity Chair, or PC Member of a number of conferences. Dr Xia has authored/co-authored one book and over 110 scientific papers in international journals and conferences. His research interests include mobile and social computing, intelligent systems, and cyber-physical systems. He is a member of IEEE, IEEE Computer Society, IEEE SMC Society, ACM, and ACM SIGMobile. Naigao Jin received his PhD from the Dalian University of Technology, China. Currently, he is a Lecturer in the School of Software, Dalian University of Technology. His research interests cover wireless sensor networks and wireless communications. Lin Feng is a professor and a PhD supervisor in the School of Innovation Experiment, Dalian University of Technology, China. His research interests cover date mining, wireless sensor networks, and Internet of Things.
doi:10.1002/dac.2332 fatcat:ueumnww56vdexich77uzcj4lli