| Efficient resource allocation is a major problem in Wireless Sensor Networks(WSNs)and in the Internet of Things(Io T)paradigm,WSNs contribute hugely by connecting all the devices for smart applications.The Io T based smart applications demand a system with intelligent processing,reliable delivery and comprehensive awareness.These requirements are hard to achieve in resource constrained WSNs due to limited battery power and the huge data involved in Io T and Industrial Io T(IIo T)applications also face the computational problems along with security challenges during communication.Therefore,in this thesis three different approaches are proposed in physical layer communication to get energy efficient,fast and secure system.The main contents of this paper are summarized as follows:1.In the first approach,energy efficient and dynamic cluster formation technique is proposed.Cluster formation is the first and the main priority for any Wireless Sensor Network(WSN).Therefore,two novel cluster formation methods are depicted to improve the system performance.The first method is the combination of Firefly algorithm and Hierarchical Maximum Likelihood(HML)for an Optical Wireless Sensor Network(OWSN).It utilizes the unique property of Firefly algorithm and overcomes the problem of it by introducing HML in it.As a result,the parameters of the network change with respect to the requirement of suitable position of nodes and the power distribution in the nodes is also accurately done based on the maximum likelihood property of HML.It means the nodes become active when they are selected based on the closest value of source node to form a cluster.It eventually shows the better performance compare to both the techniques.The second method is proposed to make the clustering process more dynamic and robust in Io T environment.The clusters are formed in this mechanism based on the power demand of the nodes and the information volume for the specific application.Initially,nodes are divided into two clusters according to OR logic,with respect to the set minimum and maximum threshold values of assigned application.Then again the nodes in the initial clusters dynamically arrange as per the information volume of new requested task in the network.The simulation results depict that the proposed method is capable of improving the energy efficiency of the network while maximizing the information volume to avoid further information loss for the desired application.These results are also compared with traditional methods to justify the proposed work.2.In case of Io T based applications,along with energy efficient clustering techniques,a proper management of power distribution can increase the network performance.Therefore,second approach deals with the problem of efficient distribution of power for smart applications.The smart applications include smart healthcare system,as a key phenomenon of new era of medical service.Smart health care system consists of different techniques and approaches,Remote health monitoring is one of those important techniques of health care system,which helps doctors to monitor health of remotely located patients with different health statistics provided by sensors.This whole process should be fast,prompt in response and energy efficient which is tough with limited battery power of sensors.Therefore,the second approach proposes a novel method for efficient utilization of power in WSNs,using Artificial Neural Network(ANN)based technique Self Organizing Map(SOM)for clustering and Distributed Artificial Intelligence(DAI)for power distribution in the nodes.The hybrid approach of SOM and multiagent based DAI results in better performance compare to other existing methods.The performance of the proposed method is validated with mathematical analysis and simulation results,which justifies the significance of the work for Io T environment.3.Io T for industrial development and the respective IIo T based applications also face the problems related to resource allocation and efficient energy utilization.The third approach deals with the problems of data security and resource allocation in IIo T network.The recent technical evolution has made its mark in industrial applications by making the network more flexible and computation friendly through connecting all the devices.As a subset of Io T,the framework of Industrial Io T(IIo T)is based on huge number of nodes with continuous process of multiple works at a time.Due to this,multi-objective network,interference in the path always becomes reason for the loss of network resources as well as the security of data becomes vulnerable.In most of the previous works,dedicated channel states are considered for fixed resources which remains a major issue of IIo T network flexibility along with security.In our third approach,both the problems are incorporated by calculating the channel security and using convolutional neural network(CNN)optimal channel state extracted for different applications.This results as a fast system with proper utilization of resources and validated with mathematical analysis and simulations. |
