Research On Tolerant Aggression Algorithms For Data Aggregation In Wireless Sensor Networks

In recent years,with the continuous improvement of related technologies of wireless sensor networks,wireless sensor networks play an important role in the fields of emergency disaster prevention,national defense military,and environmental monitoring.In these applications,wireless sensor networks are typically deployed in unmanaged,complex environments,so energy is not replenished by the outside world.And wireless sensor networks have limited energy.Due to the energy limitations of wireless sensor networks,wireless sensor network data aggregation technology has emerged.Although this technology can reduce the energy consumption of nodes in wireless sensor networks by reducing the amount of data transmitted in wireless sensor network communications.However,due to the energy and cost limitations of wireless sensor networks,nodes in wireless sensor networks are generally not equipped with tamper-proof modules and trusted computing units.This makes it very easy for malicious attackers to physically invade the wireless sensor network node,and the hacked node launches an erroneous data injection attack against the wireless sensor network data aggregation technology.Based on the working characteristics of wireless sensor networks,wireless sensor networks can only be used safely if safety problem are effectively addressed.Therefore,the data-aggregation-based tolerance algorithm in wireless sensor networks has high research value.In this paper,the traditional sensory intrusion algorithms in wireless sensor networks are researched and summarized.The two-stage algorithm is proposed based on data prediction.In most of the time,the algorithm runs in the first stage with low algorithm complexity.When only a suspected abnormality is detected,it is further confirmed by switching to the second stage with high algorithm complexity.The first stage of the algorithm is based on the characteristics of the stability of the aggregated data sequence,and the traditional algorithm is used as the prediction algorithm with the Auto Regressive and Moving Average Model(ARMA)model with low complexity.Then,the difference between the predicted value and the actual value is compared with the threshold threshold.If the difference seriously deviates from the threshold threshold,the situation is regarded as a suspected abnormality,and the second-stage algorithm is immediately activated to perform more accurate abnormal analysis and abnormality.The second stage of the algorithm is based on the characteristics of large fluctuation and non-stationary data acquisition.Combined with wavelet transform to improve the ARMA model,the improved wavelet transform-ARMA model is used as the prediction algorithm.Compared with the traditional ARMA model,the improved prediction algorithm increases the time complexity from O(p~3W)to O(p~3WNlog(N)),but the predicted mean relative error(MRE)is reduced by 24.45%,which guarantees the validity and reliability of the prediction-based two-stage intrusion algorithm proposed in this paper.This paper uses MATLAB as a simulation platform to evaluate the complexity,effectiveness and reliability of the two-stage intrusion algorithm based on data prediction proposed in this paper.Under the four simulation conditions,compared with the comparison scheme,the detection rate of the malicious injection attack in the wireless sensor network is increased by 1%~16%,the false alarm rate is decreased by 2.43%~20%,and the area under the curve(Area Under Curve(AUC))is improved by 0.008~0.11(the index is between 0.1 and 1,the closer to 1,the more effective the algorithm is).