Incremental Strategy-based Residual Regression Networks for Node Localization in Wireless Sensor Networks
Abstract:The easy scalability and low cost of range-free localization algorithms have led to their wide attention and application in node localization of wireless sensor networks. However, the existing range-free localization algorithms still have problems, such as large cumulative errors and poor localization performance. To solve these problems, an incremental strategy-based residual regression network is proposed for node localization in wireless sensor networks. The algorithm predicts the coordinates of the nodes t… Show more
“…Yan et al [14] used weighted Taylor series to further improve the estimation accuracy after positioning. Zou et al [30]established a deep learning prediction model to fine-tune the localization results according to the loss function. Ren et al [31] designed a two-stage centroid localization algorithm, which complements the single localization results of the first stage with the localization of the second stage.…”
This work proposes an improved DV-Hop model based on function analysisand simulation parameter determination (named FuncDV-Hop) to address theproblems of low positioning accuracy and strong scene dependence of the DV-Hop localization model. The undetermined coefficient optimization, step functionsegmentation experiment, weight function strategy with equivalent points, andmaximum likelihood estimation correction are introduced to reduce the positioning error by analyzing the error reasons in the average hop distance, distanceestimation, and least square calculation of the DV-Hop model. The experiment isdesigned using the control variable method. The total number of nodes, the ratioof beacon nodes, the communication radius, the number of beacon nodes, and thenumber of unknown nodes are designed as variables to control the experiment.Finally, the experiments of two stages of simulation parameter determinationand integration optimization are carried out. The simulation optimization rateof all scenes is between 23.70% and 75.76%, and the average optimization rateis 57.23%. Experimental results show that the FuncDV-Hop model has the highest optimization rate of more than 50% in all experimental scenarios comparedwith the other models, the localization error is reduced by more than 0.1, andthe optimization rate is increased by more than 10% in the record parameters ofthe existing wireless sensor network system.
“…Yan et al [14] used weighted Taylor series to further improve the estimation accuracy after positioning. Zou et al [30]established a deep learning prediction model to fine-tune the localization results according to the loss function. Ren et al [31] designed a two-stage centroid localization algorithm, which complements the single localization results of the first stage with the localization of the second stage.…”
This work proposes an improved DV-Hop model based on function analysisand simulation parameter determination (named FuncDV-Hop) to address theproblems of low positioning accuracy and strong scene dependence of the DV-Hop localization model. The undetermined coefficient optimization, step functionsegmentation experiment, weight function strategy with equivalent points, andmaximum likelihood estimation correction are introduced to reduce the positioning error by analyzing the error reasons in the average hop distance, distanceestimation, and least square calculation of the DV-Hop model. The experiment isdesigned using the control variable method. The total number of nodes, the ratioof beacon nodes, the communication radius, the number of beacon nodes, and thenumber of unknown nodes are designed as variables to control the experiment.Finally, the experiments of two stages of simulation parameter determinationand integration optimization are carried out. The simulation optimization rateof all scenes is between 23.70% and 75.76%, and the average optimization rateis 57.23%. Experimental results show that the FuncDV-Hop model has the highest optimization rate of more than 50% in all experimental scenarios comparedwith the other models, the localization error is reduced by more than 0.1, andthe optimization rate is increased by more than 10% in the record parameters ofthe existing wireless sensor network system.
“…Self-reconfigurable nodes are capable of sensing data and transmitting it to a designated sink node [1]. The wireless sensor network has various applications, such as environmental monitoring, military area surveying, traffic management, medical management, and other security purposes [2]. For these applications, the node position is crucial to detecting information which would be meaningless without location information attached [3].…”
Localization is a primary concern for wireless sensor networks as numerous applications rely on the precise position of nodes. This paper presents a precise deep learning (DL) approach for DV-Hop localization in the Internet of Things (IoT) using the whale optimization algorithm (WOA) to alleviate shortcomings of traditional DV-Hop. Our method leverages a deep neural network (DNN) to estimate distances between undetermined nodes (non-coordinated nodes) and anchor nodes (coordinated nodes) without imposing excessive costs on IoT infrastructure, while DL techniques require extensive training data for accuracy, we address this challenge by introducing a data augmentation strategy (DAS). The proposed algorithm involves creating virtual anchors strategically around real anchors, thereby generating additional training data and significantly enhancing dataset size, improving the efficacy of DNNs. Simulation findings suggest that the proposed deep learning model on DV-Hop localization outperforms other localization methods, particularly regarding positional accuracy.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.