Research On Low Energy Consumption Algorithm Of Wireless Sensor Networks Based On Machine Learning

With the development of Internet of Things technology,wireless sensor networks play an increasingly important role in production and people's daily life.The wireless sensor network has numerous and low-capable communicational senor nodes which has limited energy.It can collect the sensing object data in time and transmit it to the remote end through the network.Compared with traditional networks,the main differences is that the node energy is limited.Therefore,the methods on reducing node energy consumption and improving node energy's utilization is one of the hot issues in the current wireless sensor network field.In the current version of wireless sensor network low energy algorithm,the clustering algorithm and the low energy algorithm based on node task scheduling are two main algorithms to improve node energy utilization and reduce node energy consumption.However,the existing clustering algorithms mostly ignore the node distribution density and the node's own energy status,and it may lead to the premature death of the cluster head nodes due to unreasonable distribution or energy exhaustion;while the Q-learning-based node task scheduling algorithm can obtain a better node task scheduling strategy,but the algorithm's time and space requirements are higher.Here are three aspects of research in this paper:(1)Shortcomings of LEACH algorithm in cluster head selection,the machine learning algorithm——Mean Shift is applied to the cluster head selection process of clustering algorithm,taking into account factors such as node distribution density and node residual energy,low energy clustering algorithm(M_LEACH)based on Mean Shift was put forward.The algorithm competes for the cluster head based on the residual energy of the node and the distribution density of the surrounding nodes in each data transmission period of the network.The cluster head node collects the data of the cluster member nodes for de-redundancy processing and sends them to the base station or sink node.The simulation results of the algorithm show that compared with the direct data transmission method,LEACH algorithm and EAMMH algorithm,the M_LEACH algorithm has a better effect in solving the low energy consumption problem.(2)Insufficient time and space for the Q-learning-based node task scheduling algorithm.By modeling the modules of the sensor nodes,based on the Markov decision process model,the deep neural network is used to establish the node state and mapping relationship,which solves the problem that the traditional Q-learning algorithm needs to traverse and store a large amount of empirical data.A low-energy node task scheduling algorithm based on deep reinforcement learning is proposed.Simulation results show that for large wireless sensor networks,the deep reinforcement learning node task scheduling algorithm has a better effect on solving low energy consumption problems.