Research On Robust Localization Mechanism With Unknown Transmit Power For Ocean Sensor Networks

With the increasing contribution of the marine economy to the national economies of countries around the world,countries are paying more attention to the ocean to an unprecedented strategic height.To gain a competitive advantage,international organizations and various countries have formulated corresponding measures to deal with many challenges faced by the development of the marine economy.Our country has accordingly put forward the strategic goal of building a "maritime power with Chinese characteristics",coordinated the allocation of marine and land resources,adjusted the economic and industrial layout,and focused on the development of key marine technologies.Ocean area monitoring technology,as a key technology,has received extensive attention and vigorous development in recent years,especially the technology based on Ocean Sensor Networks(OSNs),which provides effective technical means and information services for marine ecological monitoring,marine engineering and construction,marine traffic and maritime security.In the OSNs monitoring system,the acquisition of accurate location information is crucial for military monitoring,maritime search and rescue,resource exploration,and so on.However,obtaining robust localization information in complex and dynamic marine environments is facing great challenges.On the one hand,considering that the deployed ocean nodes are generally used for a long time and are prone to aging,the transmission power has great uncertainty.In addition,the nonlinear and non-Gaussian noise caused by multiple heterogeneous sources on the water surface,natural factors such as wind and waves,or human attacks is relatively large,which increases the probability of node failure.Potential faulty nodes may result in the disconnection of the information chain between the anchor nodes and the unknown node,causing the information to deviate far from the actual value or even be lost.Moreover,the ocean link is easily affected by factors such as obstacles on the sea surface,so that the collected data is strongly interfered by the non-line-of-sight(NLOS)bias,increasing directly the difficulty of precise positioning.On the other hand,considering that the Global Positioning System(GPS)signal cannot propagate underwater,the RF signal attenuates rapidly,and the optical signal is easily scattered underwater,the acoustic signal becomes the best underwater communication signal.However,in the underwater acoustic communication channel,in addition to the high uncertainty of the transmit power,the coexistence of path loss,absorption loss,geometric expansion loss,etc.,caused by the non-homogeneous transmission medium underwater may create a rather unfavorable actual environment.To address these challenges and achieve robust localization in OSNs,in this dissertation,the respective robust localization problems are researched combined with the different characteristics of surface localization and underwater localization in the marine environment,specifically:1)To solve the problem that the positioning accuracy is not ideal due to Gaussian mixture noise caused by factors such as multiple heterogeneous sources on the water surface under unknown transmit power(UKTP),an OSNs target localization algorithm based on improved differential evolution(IDE)is proposed.The nonlinear optimization problem based on received signal strength difference(RSSD)measurements is formulated as a non-convex maximum likelihood(ML)problem.To address this,opposition-based learning(OL)theory combined with chaotic map(CM)is employed to obtain a diverse population.The adaptive mutation(AM)strategy based on two subpopulations is utilized to balance global search and convergence.The global optimal estimate can be obtained through multi-generation optimization.2)Considering the same scenario as 1)and potential fault nodes,a robust fault-tolerant localization mechanism for OSNs based on RSSD is developed.Using the first-order Taylor series approximation,the original localization problem is transformed into a non-convex optimization problem.To solve this,the initial estimate searching method(IESM)is used,which relies on the active set method.In order to accelerate the convergence and enhance the robustness,a regularization item(RI)is introduced,and the optimal solution is obtained in multiple iterations through the block-update surrogate function(BUSF).3)For the adverse effects of non-line-of-sight(NLOS)bias on real-time multi-target positioning when the transmission power is unknown,and considering the highly non-convexity of the problem,a fast multi-target localization technique based on neural network(FMLNN)is presented.Offline training is performed with a prior dataset consisting of known target information and RSSD measurements to learn nonlinear mappings.After the training is accomplished,the FMLNN with the optimized structures can obtain estimates of multiple targets online and in real time,alleviating the NLOS effect to achieve effective multi-target localization.4)To alleviate the double attenuation of underwater path loss and absorption loss under UKTP and realize the joint estimation of target information and transmit power in OSNs,a robust two-stage underwater acoustic localization scheme in OSNs is proposed.The original received signal strength(RSS)-based localization problem is converted to the unknown transmitting-power alternating non-negativity-constrained least square(UT-ANLS)framework,which is solved in two stages.In the first stage,the interior point solving method(IPSM)is used,which mainly uses the logarithmic barrier function(BF)to limit the optimization bound to obtain a rough estimate.In the second stage,the target position and transmit power are further tuned by the majorization-minimization tactic(MMT)to improve the accuracy of underwater localization.Research will help improve the ability to cope with additional effects caused by various unfavorable factors in the case of unknown transmit power in OSNs,and provide accurate location data for applications such as deep-sea data collection,intelligent ship navigation,and moving target tracking in OSNs,so as to ensure the smooth progress of various activities without resistance.In addition,the research has a good reference for the application of localization in wireless sensor networks(WSNs)under other complex backgrounds,is of great significance for promoting the development of robust and fault-tolerant positioning in WSNs,and has broader application prospects for future marine military or civil engineering.