Research On Congestion Control Methods Of Wireless Sensor Network Based On Node Queue Management

In recent years,the technology of Wireless Sensor Network(WSN)has been developing rapidly,and the application fields are more and more widely.Congestion in WSN can be divided into multiple levels,one of which is node level congestion,that is,the packet traffic that nodes need to send exceed the transmission capacity,resulting in buffer queue overflow,data packets loss and network queuing delay increase.As a result,the network throughput is reduced,and the whole network will be paralyzed when it is serious.Therefore,it is necessary to study a network congestion control method which can effectively mitigate the level of wireless routing nodes.The main research content of this paper is to solve the network congestion caused by wireless routing nodes forwarding a large number of data in WSN.At present,the node queue management method of WSN can alleviate the congestion problem caused by this reason.Active queue management(AQM)is an important research idea in congestion control of wired network,and this idea has been introduced into WSN.The AQM method can loss packets before congestion.It can control queue length effectively,avoid network congestion,and solve the problem of unnecessary consumption of lag and energy.Therefore,this paper studies the congestion control method of WSN based on node queue management.Firstly,this paper introduces the WSN and WSN congestion control,discuss WSN congestion control methods.The PI algorithm is analyzed in detail,and the PI algorithm and the Droptail algorithm used in WSN are simulated and compared.By analyzing and summarizing the advantages and disadvantages of the PI control ideas applied to each wireless node in the queue management of the WSN.Secondly,in order to solve the problem of fixed parameters of PI algorithm,this paper proposes a Particle Swarm-Single Neuron-PI(PNPI).The PNPI algorithm uses neuron's self-learning and self-organizing ability,and optimize the selection of parameters by adjusting the PI controller's proportional and integral weight parameters online.Then the standard particle swarm optimization(SPSO)algorithm is used to optimize the learning rate of neuron in the neuron PI(NPI)algorithm.Through the real-time correction of the learning rate of neuron,the weights of single neuron can be adjusted online to prevent the local optimization of neuron algorithm.Finally,before the node queue buffer overflows,drop packets with p obtained by the PNPI algorithm to avoid network congestion.Through the simulation experiment,the PNPI algorithm improves the performance index of packet loss rate,throughput and delay.Afterwards,considering that the PI controller has no differential regulation,the PID control technology is applied to the cache queue of the WSN nodes.Its purpose is to accelerate the adjustment time of the algorithm,effectively control the stability of the buffer queue length of the WSN nodes,and improve the network performance.In this paper,a congestion control method Particle Swarm-Single Neuron-PID(PNPID)is proposed.First of all,the PID control idea is applied to the queue management of WSN nodes.Then use the neuron control technology to adjust the proportional,integral and derivative parameters of the PID controller online.Afterwards,the SPSO algorithm is used to perform on-line optimization of the initial values of the proportional,integral,differential parameters and neuron learning rates in the neuron PID(NPID)algorithm.An improved PID algorithm is designed through TCL language and C++ programming language.The improved PID algorithm is loaded into NS2,and different simulation environment is applied to simulate the improved algorithm.The simulation results show that the PNPID algorithm can well stabilize the queue length near the expected value,and at the same time,the network performance indicators such as throughput and packet loss rate have also been greatly improved.The network congestion is relieved and the network QoS is improved.Finally,the PNPI and PNPID algorithms proposed in this paper are compared.It is found that the packet loss rate and throughput performance of the PNPID algorithm are better than that of the PNPI algorithm.Finally,the shortcomings of the improved PNPID algorithm are pointed out,and the next work plan of this paper is prospected.