Research On Key Technologies For Efficient Transmission And Intelligent Processing Of Ocean Sensor Networks Data

Ocean Sensor Networks(OSNs)have been widely used in ocean environmental monitoring,maritime search and rescue(MSR),and marine information platform construction with their advantages of self-organization,self-adaptation,low-power consumption and strong scalability.The strategic needs of national maritime power and the growing demand for human exploitation of marine resources and marine economy make it urgent to establish a global ocean monitoring and MSR system,and the development of observation technologies such as OSNs and advances in information technology make it possible.In the OSNs,intelligent data processing is its key technology in ocean monitoring and MSR applications.This paper mainly studies four key technologies in the intelligent data processing of OSNs,including opportunistic routing protocol(efficient data transmission),missing data recovery,target detection,and multi-target localization and tracking.Among them,efficient data transmission and missing data recovery provide complete data support for the dynamic estimation of ocean targets.The main challenges facing the research of intelligent data processing in OSNs are as follows: 1)The sensor nodes deployed on the sea move in real-time under the action of wind,waves and currents,that is,the network topology is highly dynamic,which makes it difficult to form a stable dynamic routing path;2)Due to the high dynamics of the network topology and the influence of the wave shadowing effect,ocean communication links have the characteristics of low reliability,high probability of interruption,and strong time-varying characteristics,which makes node data more easily lost;3)The energy of ocean nodes is limited,and charging and replacement are usually impossible.Therefore,the collected data is directly transmitted to the sink node without processing,which will cause a lot of waste of transmission resources,and the life cycle of the network will be greatly affected;4)Under complex sea conditions,ocean nodes are vulnerable to Byzantine attacks and become Byzantine nodes,and nodes need to know their location information to provide effective data to the data fusion centre,so when OSNs are attacked by Byzantine nodes,how to achieve the precise localization and tracking of ocean mobile targets is a difficult problem in the field of signal processing.In response to the above-mentioned challenges,and around the key technologies of intelligent data processing in OSNs,this paper has carried out the following four aspects of work:1)Efficient transmission of ocean data: Aiming at the characteristics of ocean sensor network routing design,an energy-efficient opportunistic routing protocol based on compressed sensing and power control mechanism is proposed.To reduce the frequent exchange of location information between nodes and its neighbors,the temporal and spatial relationships of OSNs network topology are used and combined with the weighted moving average method to predict the distance of data packets.Subsequently,an adaptive power control mechanism was proposed to select the optimal transmission power and candidate forwarding node set.In addition,the compressed sensing technology greatly reduces the amount of data collected by ocean nodes and the amount of data transmitted in the network.Finally,in order to avoid packet conflicts,a timer-based candidate node set scheduling algorithm is utilized to coordinate data packet forwarding.The simulation results show that,compared with the existing algorithms,the proposed opportunistic routing protocol of the OSNs effectively improves the packet delivery rate,reduces network energy consumption and end-to-end delay.2)Ocean missing data recovery: Considering that the data collected by OSNs nodes has strong spatio-temporal correlation,a node missing data recovery algorithm KPR-NN based on improved K-means algorithm and APSO-RBFNN(Radial Basis Function Neural Network Optimized by Adaptive Particle Swarm Optimization Algorithm)is proposed.KPR-NN includes node clustering module and data recovery module.First,use the improved K-means algorithm in the node clustering module to cluster nodes into clusters.Then,in the data estimation module,the APSO-RBFNN is utilized to accurately recover the missing data.Under the condition that the data error threshold is met,KPR-NN can effectively reduce communication cost and prolong the lifetime of OSNs by using the predicted values as a replacement for the real values in cluster head nodes.Analysis and simulation results demonstrate that,compared with the benchmark algorithms,the KPR-NN algorithm can accurately recover the missing data of ocean nodes.3)Ocean target detection: Firstly,the target monitoring sea area range is determined and a node adaptive clustering mechanism is designed to establish the OSNs clustering topology structure.Aiming at the problems of low accuracy and poor robustness of existing ocean mobile target detection algorithms,a mobile target detection algorithm based on KL divergence and information gain is proposed.KL divergence is adopted to measure the strength of wave shadow effect and reduce the False Alarm probabilities of target detection.In addition,an adaptive dynamic decision threshold is derived based on relative entropy and global optimal decision statistics to ensure the accuracy of target detection.To conserve the overall OSNs energy and improve detection efficiency,cluster with the maximum expected information gain are selected as target search clusters before the next round search.The simulation results verify that the proposed algorithm is suitable for highly dynamic and highly noisy ocean environment,and Detection/Flase Alarm probabilities of target detection are better than the existing algorithm.4)Ocean multi-target localization and tracking: In view of the situation that OSNs nodes are vulnerable to Byzantine attacks,firstly,an efficient Byzantine node identification method based on dynamic threshold is presented by using the information entropy of a system composed of a single sensor and their neighbor sensors.Ocean target localization and tracking problem is formulated using both received signal strength indication(RSSI)and Maximum likelihood(ML)frame after migrating Byzantine nodes.Then,the Levenberg-Marquardt algorithm is utilized to solve the ML localization problem by combining the prior location information and uncertainty of the beacon/honest nodes.The proposed robust distributed cooperative multi-target localization and tracking algorithm employ the importance sampling method to approximate the posterior probability distribution of sensors and targets location.Furthermore,when the marine target is outside OSNs coverage,a piecewise function is adopted to characterize the probability of finding falling water target in the search and rescue sea area.The results of analysis and simulation show that although the proposed algorithm is not optimal in execution time performance,the localization and tracking performance is better than the benchmark algorithms at low beacon node density and high measured noise.The research results have positive guiding significance for promoting the application of wireless sensor networks in marine environment monitoring and MSR.At the same time,they also have important scientific significance and academic value in related theories such as the dynamic acquisition and intelligent processing of ocean big data.