Research On Key Technologies Of Precision Electronic Interference Based On Wireless Sensor Networks

Because that the wireless sensor network(WSN)has flexible topological structure,it is highly adaptive to complex and changing battlefield environments.Thus,WSN tends to be developed as platforms to implement precision electronic interference,which has recently become a new research focus in the field of electronic warfare.However,the large sensor number in WSN considerably increases the processing complexity and there-fore leads to a huge challenge.To address this obstacle,we first design a highly efficient precision interference system.Then,we propose cooperative localization algorithms with distributed computation framework and precise interference algorithms.The major con-tributions and novelty are illustrated as follow:1.We establish a distributed precision interference framework based on WSNs.This framework is designed according to practical requirements.In this framework,we first estimate the positions of all sensor nodes via cooperative localization,and then build the ultra-sparse array according to the prior information of positions.Afterwards,we design optimal array transmit signals to focus interference energy on target areas and suppress delivered energy on protected areas below a safety level.Depending on this framework,we also propose several metrics to evaluate the performance of cooperative localization and Focused Energy Delivery(FED),which are considered and used in the subsequent research.2.In order to tackle the two major challenges of cooperative localization,which are low computational efficiency and severe accuracy loss caused by lacking prior knowledge,we respectively propose three distributed cooperative localization algorithms with high computational efficiency according to three different scenarios.(1)We propose a distributed cooperative localization algorithm for Line-of-Sight(LOS)environments.Traditional cooperative localization techniques in a centralized framework have considerably high computational complexity especially when the number of sensor nodes increases.Therefore,we propose a distributed algorithm framework to address this drawback.Firstly,we design a redundant model to project the indecompos-able localization optimization problem onto exclusive high-dimensional spaces and make the original problem decomposable in expression.Secondly,we use Alternating Direc-tion Method of Multipliers(ADMM)to decompose the original problem into numerous small-scale sub-problems.This parallel solving framework for sub-problems breaks the scale limit of WSN,and significantly decreases the complexity through decomposing and dimension reduction.Thirdly,when solving sub-problems,we relax the non-convex sub-problems into Semi-definite program(SDP).Finally,we propose an adaptive model of penalty parameters to guarantee convergence of ADMM for the non-convex problem.Theoretical analysis and simulation results show that the proposed algorithm successfully decreases computational complexity and has higher accuracy than stat-of-art algorithms.(2)We propose a distributed cooperative localization algorithm for Non-Line-of-Sight(NLOS)environments.In the NLOS scenario,distance measurement error leads localization accuracy to degrade severely,and the centralized processing framework al-so causes low computational efficiency.We thus propose a distributed cooperative lo-calization method for WSNs to mitigate NLOS impact.Firstly,we establish a heuristic modification model based on the range measurement.This model is applicable when lack-ing prior knowledge about NLOS connections.Secondly,we use projection relaxation to convert the proposed non-convex model into its convex envelope.Thirdly,we developed an efficient decomposed formulation for the convex counterpart,and designed a parallel distributed algorithm based on the ADMM,which significantly improves computational speed.Finally,to accelerate the convergence rate of local updates,we approach the sub-problems via the proximal algorithm.Numerical simulation results demonstrate that our method is superior in processing speed and accuracy to other methods in NLOS scenarios.(3)We propose a cooperative localization algorithm for LOS/NLOS environments that achieves both high accuracy and robustness.Existing algorithms are only designed specially for LOS or NLOS environments,which means they are not general for situa-tions without prior knowledge about LOS/NLOS connections.To solve this problem,we propose a general cooperative localization algorithm that suits LOS/NLOS simultaneous-ly.Firstly,we introduce modification coefficient to build a highly robust multiplicative model and use projection relaxation to convert the non-convex problem into its convex envelope.Secondly,we propose a parallel distributed framework to decompose the large-scale problem.In the framework,we derive and propose proximal method to solve the non-smooth sub-problems.Finally,we analyze and derive several significant metrics of the algorithm which evaluate the convexity,convergence and computational complexity.Simulation results and theoretical derivation both show that our algorithm effectively re-move the barrier between LOS and NLOS cooperative localization and achieves both high computational efficiency and accuracy compared to existing algorithms.3.Because that existing algorithms designed for precision interference generally have considerably high computational complexity and that ultra-sparse arrays will lead to severe grating lobe effect,we propose three efficient focused energy delivery(FED)algorithms.(1)We analyze the feasibility of precision interference based on WSNs.Firstly,We derive the Cramer-Rao Lower Bound(CRLB)of cooperative localization and analyze the factors that can influence localization accuracy.Then,we propose the necessary error bounds that satisfy FED requirements.Theoretical analysis and simulations results show that accuracy of cooperative localization can satisfy the basic requirement of FED in com-mon communication frequency bands and some radar bands.(2)To decrease the computational burden,we propose an FED algorithm based on the majorization-minimization(MM)method.We directly optimize the transmit signals of the array and ensure the energy to be focused on target areas and to be suppressed on protected areas via building a QCQP model.Then,to obtain the superior solution,we build the tight upper bound of the original quadratic objective function,and adopt this upper bound as the surrogate function in the iterative optimization process.Theoretical analysis and simulation results both demonstrate that our algorithm has much lower computational complexity when compared with other state-of-art algorithms.(3)To mitigate the grating lobe effect,we propose a fast FED algorithm based on M-M and a distributed FED algorithm.Firstly,we introduce both the FED metric and grating lobe metric to rebuild the FED optimization model.Secondly,to solve the optimization problem that contains both quadratic term and l₁norm,we derive the upper bound function of the new objective function and solve the problem fast by using MM twice.However,this fast algorithm still can be improved because the eigen decomposition in the second MM method brings additional complexity.Thus,we propose a distributed FED algorith-m.Firstly,we introduce interactive variables to optimize the quadratic term and l₁norm in an alternating and independent way.Secondly,we use MM and proximal method to solve QCQP problems and l₁norm problems respectively,which further decreases the complexity.Simulation results show that the fast FED and distributed FED algorithms we propose both suppress the grating lobe,and the distributed FED algorithm has much lower complexity than fast FED algorithm.