Research On Real-time Graph Routing Algorithm Based On SDWSN

A software-defined architecture is introduced into the wireless sensor network to realize the separation of control and data forwarding planes,reduce operating costs,and prolong network life.Based on this,this thesis studies the real-time graph routing algorithm based on software-defined wireless sensor networks.Combined with a routing algorithm that takes both energy efficiency and detection reliability into consideration,a software-defined wireless sensor network routing algorithm based on reinforcement learning is proposed,which can prolong the network life and achieve the goal of improving network performance detection reliability and network energy utilization.The main work is as follows:1.Aiming at the distribution of the number of cluster head nodes in SDWSN: This thesis analyzes the characteristics of network data transmission,designs a network energy consumption rate estimation model,which combines the location,density and transmission delay of nodes.And the energy consumption of nodes is calculated by this model.Analyze the energy consumption of hot spots and non-hot spots to determine the total number of cluster heads;Then,the thesis designs a cluster head allocation principle according to the energy consumption rate in different event 1areas: the energy consumption rate of the event area is inversely proportional to the number of cluster heads.Finally,the energy utilization rate of the low-consumption area is improved,and the energy consumption of the high-energy consumption area is reduced,so as to balance the network energy consumption.Since more cluster heads are selected and allocated according to the energy consumption rate,the event detection reliability is improved.2.Aiming at the problem of cluster head selection: The thesis designs a cluster head selection algorithm based on dynamic radius.The algorithm selects cluster head by dynamically adjusting competition radius according to density weight,distance weight and energy weight.And introducing assistant cluster head as forwarding node of the cluster heads.Finally,the energy consumption of cluster head data transmission is reduced,the network energy consumption is balanced,and the network life is prolonged.3.Aiming at the problem of routing path selection: This thesis designs an optimal path selection algorithm based on RL.The transmission success rate of data packets,the remaining energy ratio of the cluster head,the ratio of the distance from the adjacent node to the sink node to the maximum distance to the sink node,the ratio of the number of hops from the adjacent node to the sink node and the maximum number of hops to the sink node are added to the reward function.Finally,the pros and cons of the current path are accurately evaluated,and the optimal path is selected,thereby prolonging the network life.Finally,this thesis compares the SCMR,RMER,EASDN and RLBR algorithms with the SDWSNRRL algorithm,and uses the OMNe T++ simulation software to simulate the five algorithms in the same environment.First,thesis analyzes the network life cycle:The network lifetime of SDWSNRRL algorithm is increased by 4.80 times,1.15 times,23.4% and 9.1% compared with SCMR,RMER,EASDN and RLBR algorithms,respectively;And when the learning rate changed from 0.3 to 0.5,the network lifetime of SDWSNRRL,EASDN and RLBR algorithms all increased by 3% to 5%.Secondly,the Packet Drop Rate(PDR)is analyzed and it is found that the PDR of SDWSNRRL algorithm is 0.94 times,0.87 times,0.38 times and 0.92 times lower than that of SCMR,RMER,EASDN and RLBR algorithms,respectively.And the analysis of the network distortion degree shows that the average detection distortion of SDWSNRRL algorithm is0.21 times and 0.09 times lower than that of SCMR and RMER algorithms,respectively.SDWSNRRL network energy utilization is 8.12 times and 1.28 times higher than that of SCMR and RMER algorithms,which improves the balance of network energy consumption.