Optimization Model And Research Of Energy Allocation Based On T-AMC In Energy Harvesting Internet Of Things

In the Energy Harvesting(EH)type Internet of Things system(Io T),due to the limitation of available energy and spectrum resources,how to reduce the number of data transmissions(or transmission delay)on the premise of accurately reconstructing the signal has always been a very challenging problem.Compressive Sensing(CS),as a new data processing technology,does not need to meet the Nyquist sampling theorem,and can accurately reconstruct the signal through a small amount of sampling,which can effectively reduce the number of transmissions of the system,and reduce the cost in the process of signal processing and transmission.In the EH-Io T,each EH node maintains its own operation by collecting the surrounding energy,and consuming a certain amount of energy to transmit its observation value to the fusion center(FC)through the multiple access channel(MAC)communication model,after the (i.e.,the number of transmissions)time slot,the FC receives the compressed version of the original signal.In this scenario,the energy allocation(or energy consumption)of all sensor nodes in each time slot is coupled with the channel effect,which is equivalent to the measurement matrix in the standard CS theory.The performance of this measurement matrix determines the requirements of the FC for accurately reconstructing the original signal for the number of transmissions.A measurement matrix with good performance can have better signal reconstruction quality at the same number of transmissions .The research target of this thesis is to optimize the energy allocation of each time slot and construct a measurement matrix with excellent performance under the dynamic condition of available energy constraints in EH-Io T.The main research contents and innovations are as follows:(1)Based on the MAC communication model and CS theory,this thesis transforms the energy allocation problem in EH-Io T into the construction of the measurement matrix.By analyzing and utilizing the t-average mutual coherence(t-AMC)and mutual coherence(MC),under the condition of satisfying the available energy constraints,a new energy optimization allocation strategy OA-t-AMC and OA-MC are proposed,and the corresponding online model is given on this basis.Through numerical experiments,it is proved that the signal recovery quality of the proposed energy optimization allocation scheme is compared with the traditional The strategy based on probability allocation has been significantly improved.(2)Due to the high time complexity of solving optimization problems either through convex optimization or Newton-like methods,the OA-t-AMC and OA-MC optimization algorithms may not be applicable in systems with low time delay requirements.Therefore,this thesis uses neural networks to replace this construction process.The training data set is generated by the OA-t-AMC algorithm,and an OA-t-AMC-CNN network model based on convolutional neural networks(CNN)is proposed,and an adjustment strategy is proposed to make the output through the neural network meet the constraints,and online model is also given.The output of the neural networks meets the constraints.Numerical experiments have proved that the OA-t-AMC-CNN model can achieve similar performance as the OA-t-AMC model,but the time required for energy distribution is greatly reduced.