Research On Localization Technology Of Wireless Sensor Network In Intelligent Plant

With the rapid development of sensor technology,wireless communication technology,embedding technology,distributed processing technology and MEMS technology,Wireless Sensor Network has appeared and it is widely applied in Smart Medical,Intelligent Transportation,Military Reconnaissance,Intelligent Home Living,Green Environmental Protection,Intelligent Industry and etc.Wireless Sensor Network is composed of a large quantity of sensors that constitute a self-organized,scalable,point-to-point communication,multi-hop transmission wireless network.Through environment perception,information fusion and data transmission,a large number of valuable information is transmitted to the observer.In the applications of wireless sensor networks,sensor information includes not only the perceptual data,but also the specific location of the data.Perception data without detailed location information is meaningless,so Node Localization technology is one of the most important and fundamental technologies in Wireless Sensor Network.Wireless Sensor Network based ubiquitous sensing computing network is an important sensing and monitoring infrastructure of Intelligent Plant.In Intelligent Plant,the reason that the decision maker or the intelligent analysis system can accurately analyze the current situation and take the next effective measures is that they can perceive and collect a large number of important parameter information such as environment,workers,equipment,energy,etc.Most of the sensing data is location-aware,so in the intelligent plant,the localization and tracking of AGV,the localization of workers and outsiders and the locations of characteristic parameters or events in the monitored area are all inseparable from wireless sensor network localization technology.According to the applications required in different situations,the localization algorithms mainly includes: static and dynamic localization,ranging-based and ranging-free localization,single target and multi-target localization.In addition,the localization algorithm should meet the requirements of fast response speed and high localization accuracy.By researching on many domestic and foreign literatures relevant with wireless sensor network localization technology,the current widely recognized wireless sensor network localization technologies are mastered and studied how to raise the localization accuracy,reduce the response time and improve the robustness to be more adaptable in the intelligent plants.The main research and improvement are as follows:(1)Range-free Distance Vector Hop(DV-Hop)localization algorithm is researched.Considering that DV-Hop has large localization error,a novel DV-Hop localization algorithm based on Common Error Vector Modification and a novel DVHop localization algorithm based on Fuzzy Clustering Error Vector Modification are proposed in this paper.The improved DV-Hop constructs the similarity function according to the network topology structure(WSN nodes single-regionally deployed and WSN nodes regionally deployed),uses the minimum hop count between the anchor nodes and the unknown node to get the similarity values and uses the predefined similarity threshold to select the set of anchor nodes that have a certain similarity with the unknown node.The anchor nodes in the set are re-localized to get the estimated locations of them and then according to the principle of Affine Space,we get the error vectors or modification vectors of the anchor nodes relying on their estimated and real locations.Finally,the location error vector or modification vector of the unknown node is calculated as the weighted average value of the error vectors or the modification vectors of the anchor nodes to further modify the estimated location of the unknown node.Experiments show the improved DV-Hop algorithm can effectively raise the localization accuracy of range-free DV-Hop algorithm in different WSN topology.(2)Research of the Range-Based Trilateration Weighted Centroid localization algorithm.Because the localization accuracy of the RSSI ranging based localization algorithm is greatly influenced by signal noise and environment disturbance,three aspects of improvements of RSSI ranging based localization algorithm are proposed:Firstly,to resolve the problem that the improper selection of anchor nodes in the traditional trilateration weighted centroid localization algorithm leads easily to large localization error,the Double-Set Combination method is proposed to search for three appropriate anchor nodes that satisfy the localization condition for trilateration localization.Secondly,considering that RSSI based ranging algorithm is prone to be interfered by events of small probability and large interference of environmental noise or equipment pulse noise,which will reduce the accuracy of RSSI based ranging result,a Fuzzy C-Mean Clustering filtering algorithm combined with Quantum Particle Swarm optimization is proposed to eliminate polluted RSSI of small probability and large interference event.Then the correctness of the filtered RSSI triplets is tested according to the Hypothesis Test method.Finally,Reference Point Weighted Centroid Localization algorithm is proposed to address the problem that the traditional trilateration weighted centroid localization algorithm has not high success rate and large error under environment noise and radio pulse interference.The effectiveness of the improved trilateration localization algorithm is verified and demonstrated by using CC2530 transmission module and Tiny OS2 development platform.(3)Monte Carlo Box Localization algorithm for mobile wireless sensor network is studied.Monte Carlo Box Location algorithm is an anchor nodes information constrained Monte Carlo sampling Location algorithm,which is commonly used for mobile target localization and tracing in Intelligent Plant.In order to solve the problems of low sampling efficiency,too many iterations,sample degrading and random sampling of Monte Carlo Box localization algorithm,in the sampling phase of traditional Monte Carlo Box localization algorithm,the Grey Model Prediction based sampling scheme is added according to the Grey System Prediction theory.The random sampling process in the sample box is improved to predict and sample current locations according to the previous sample positions.Therefore,the sampling prediction is more purposeful,targeted and directional,which greatly improves the sampling efficiency and reduces the number of iterations.Simulation results show that the improved algorithm can effectively prevent samples from degrading with the time periods passing,significantly improve the localization accuracy of the mobile target in Intelligent Plant and reduce the localization response time.(4)Research of Compressed Sensing based sparse multiple target localization algorithm.In order to address the issues of large dimension of sensing matrix,high computation complexity and difficulty in refining grid's side length in traditional Compressed Sensing based multiple target localization algorithm,a novel two-phase multiple target localization algorithm based on Compressed Sensing is proposed.In the phase of large-scale localization,the monitored area is divided by Voronoi diagrams according to the deployment of the sensors and the targets are determined in the subareas.The targets' sparse location vector is reconstructed by Greedy Matching Pursuit algorithm in the subareas and the candidate grids containing the target are obtained.In the fine localization phase,the candidate grids are refined according to the Least Grid Side Length theorem.In the local areas composed of the candidate grids in each subarea,the 1-sparse location vector Greedy Matching Pursuit reconstruction algorithm is used to get the refined girds containing the targets and then the refined grids are taken as the final locations of the targets.The two-phase multiple target localization algorithm based on Compressed Sensing not only improves the localization accuracy,but also reduces the dimension of sparse reconstruction sensing matrix,the algorithm complexity and the localization response time.Finally,the effectiveness of the proposed algorithm is verified by experimental simulation.The improved algorithm can localize the monitored sparse multiple targets effectively in Intelligent Plant.(5)Research of Multi-Dimensional Scaling based collaborative multiple target localization algorithm.To address the problem that the localization accuracy of Wireless Sensor Network Multi-Dimensional Scaling localization algorithm is easily influenced by the distance error of the shortest path estimation and to modify the initial estimation locations of the sensor nodes,Multi-Dimensional Scaling Localization Algorithm based on Simulated Annealing Optimization is proposed.RSSI is used for ranging and establishing the distance matrix.Then the initial estimated absolute coordinates of the sensor nodes are calculated by Multi-Dimensional Scaling method combined with the anchor nodes' information.The stress coefficient of Simulated Annealing Optimization is obtained from the ranging values and the weight coefficient of the nodes.The optimum locations of the sensor nodes are calculated by the iterative optimization process thanks to the excellent global search ability of Simulated Annealing Optimization.The simulations demonstrate that even in the circumstance of C-Type network topology and ranging noise Multi-Dimensional Scaling Localization Algorithm based on Simulated Annealing Optimization has lower average localization error and higher network coverage ratio,meanwhile the improved algorithm in very adapted to be applied in Intelligent Plant where the number of base-station is small.This paper systematically studies the localization technologies of wireless sensor network,which are associated with range-based and range-free localization,static and dynamic localization,single target and multiple target localization,puts forward five innovative and improved schemes and verifies the effectiveness and feasibility of the proposed algorithms through a large number of experiments and simulations.The experiment results show that the five proposed localization schemes are important innovations and improvements of wireless sensor network localization technology and can be applied in a variety of localization fields of wireless sensor network of Intelligent Plant and have very important practical significance and application value.