Research On Safe Charging Tasks Scheduling In Wireless Rechargeable Sensor Networks

Recently,Wireless Power Transfer(WPT)technology is undergoing rapid devel-opment due to its advantages such as no contact,no wiring,reliable power supply,and ease of maintenance.After the rapid development of recent years,wireless recharge-able sensor networks have been widely used in many fields,such as medical care,environmental detection and smart grid.Charging utility optimization problems in Wireless Rechargeable Sensor Net-works focus on how to design charging schemes for chargers and schedule sensor nodes appropriately to maximize the performance of the network.The existing works focus on for the scenarios where all chargers are static designing deployment scheme with chargers' adjusting factors unchanged by time,selecting paths for the scenarios where all chargers are movable,however overlooking the possibility to schedule chargers for static chargers.This thesis studies the problem of safe charging tasks scheduling in Wireless Rechargeable Sensor Networks,which is about how to schedule static charg-ers with the optimization goal of maximizing the effective charging energy and further minimizing the charging time,as well as guaranteeing the Electromagnetic Radiation Safety.This thesis is the first to study and propose a scheduling scheme for safe charg-ing wireless charging tasks.A centralized algorithm for this problem as well as a distributed algorithm that is scalable with network size are proposed in this thesis.This thesis first focuses on maximizing the effective charging energy,and for the target of reducing the charging time,proposes a centralized algorithm for small scale sensor networks and a distributed algorithm for large scale sensor networks on the basis of maximum the effective charging energy.To maximize the effective charging energy,this thesis applies an area discretization technique to approximate the continuous and nonlinear EMR safety constraint as a finite number of linear constraints.Then,this the-sis proposes an approach called solution regularization to map any arbitrary solution to a piecewise constant solution without performance loss,and transforms the prob-lem to a linear programming problem that can be easily addressed.For the centralized version to reduce charging time,this thesis proposes two algorithms,Quadratically Constrained Linear Programming based Greedy Test Algorithm and Linear Program-ming based Greedy Test Algorithm.The former is too complex to solve,but the latter not only yields the optimal result but also has fast convergence speed,which is only related to the accuracy of solution and the size of linear programming problems.For the distributed version to reduce charging time,this thesis first proposes an area parti-tion scheme to partition the whole area into many subareas,and then,it can be safely considered that each subarea independently and in parallel.With reasonable size of subareas,the global solution can be calculated using the solutions of subareas.Further,to bind the overall effective charging energy and charging time,two approaches called area-scaling and EMR-scaling are proposed.The former is to decouple the complex re-lationship between the achieved effective charging energy and charging time,the latter is to artificially adjust the EMR constraints for the global solution.This thesis proves that the ultimate solution is feasible,and achieves 1/(1 + ?)optimal in centralized al-gorithm.It also can receive overall effective charging energy no less than(1-?)of that of the optimal solution and charging time no more than that of the optimal solution.To validate the performance of algorithms proposed in this thesis,both simulation and field experiments are conducted.The results of simulation experiments show that the distributed algorithm proposed achieves 94.9%of the optimal effective charging energy and requires 47.1%smaller charging time compared to the optimal one when? ? 0.2.Furthermore the results of field experiments show that the algorithm proposed outperforms the comparison algorithms in terms of both effective charging energy and charging time.