Research On Node-level Congestion Control Based On NPID In Wirelsess Sensor Network

The application fields of wireless sensor networks are gradually expanding,and many scholars are constantly studying the key technologies involved in wireless sensor networks.Congestion control is one of the main research directions.Due to the wide application of Io T technology,massive data are being generated all the time.However,increasing the network resources blindly can not fundamentally solve the problem at all.Therefore,we must design a complete set of congestion control mechanisms for WSN to solve this question.This paper chose node-level congestion control in WSN as the research object,through the analysis of the characteristics of various algorithms,chose the AQM algoritnm based on control theory,NPID algorithm as the research basis.As a mature control algorithm,NPID has a good control effect,such as clear control objectives,scientific superiority and good robustness,which is a big step in the improvement of PID algorithm.However,through the analysis of its principle,we find that NPID algorithm is not so perfect as we consider,and it also has shortcomings.On the one hand,in this paper,aiming at the single neuron fixed gain problem of the NPID algorithm,the corresponding improvement was made,and the FNPID algorithm was proposed.Since fuzzy control does not depend on the model of the control object and has strong adaptability to uncertain models,this paper introduced fuzzy control.The queue length error e and the error rate of change ec were used as the input of the fuzzy controller.The output of the fuzzy controller,△K,was obtained through the real-time feedback of the system,so that the gain K of the single neuron could be adjusted in real time,and the intensity of the algorithm’s control on the nodes was adjusted.First,the effectiveness of the algorithm was verified by the experimental simulation experiment of the fixed topology of NS2,and then the robustness of the algorithm was verified by the simulation experiment of the random topology.The experiment shows that the FNPID algorithm maintains the length of the node queue and reduces the network transmission delay,reduces packet loss rate,and improves throughput.On the other hand,this paper conducted an in-depth study on the NPID algorithm and found that the learning rateη₁、η₂、η₃ of the single neuron and the setting of the initial parameters K_p0、K_i0、K_d0 of the PID controller were also important.Combining the characteristics of the current swarm intelligence algorithm,the dragonfly algorithm was added to optimize the above-mentioned parameters in real time,and the single neuron learning rate was adjusted in real time to achieve the purpose of single neuron weight correction,which could avoid the local optimization problem caused by the single neuron algorithm.Therefore,the most suitable parameters for the network environment to improve the performance of the algorithm could be found.Finally,a simulation experiment is carried out through NS2 to verify the effectiveness of the DA-NPID algorithm.