Network Resource Management Technologies For Industrial Internet Of Things

Industrial Internet of Things(IIoT)is the product of manufacturing industry that is deeply integrated with the new generation of information technologies such as big data,edge computing,and wireless power transfer.It aims at.connecting all sectors of industry,including design,manufacturing,management,and service,with the ultimate goal of achieving intelligent sense,ubiquitous connectivity,precise control,digital modeling,real-time analysis,and iterative optimization.Productive as IIoT is,the application of new technologies also sets new requirements for system communication.On the one hand,IIoT needs to support the interconnection of massive devices,facilitating their sustainability and meeting their large-scale data processing demand.On the other hand,IIoT also has to consider data freshness,signaling transmission costs,and stability problems in the system.Those emerging requirements pose great challenges to network resource management in IIoT.This dissertation makes an exploratory study on the network resource management technologies for IIoT,and mainly focuses on the following three challenges:1)Device side:the dense deployment and heterogeneity of IIoT greatly increase the energy consumption of cell search of wireless devices,whose limited battery capacity is unable to support their long-term operation;2)Access side:the deep coupling of communication,energy,and computing resources severely restricts the efficient usage of wireless resources;3)Network side:the interdependence between sub-networks of IIoT results in system vulnerability,leading to the increase of large-scale failures.The main contribution and innovations of the dissertation are summarized as follows.First,an inter-frequency cell search strategy on the device side of IIoT is proposed to solve the problem of surging energy consumption caused by dense deployment and heterogeneity of the network.It can reduce the energy consumption of unnecessary inter-frequency scanning by estimating whether the position of a device is within a wide-band small cell.Specifically,a concept named inter-frequency measurement detection region is proposed,which guides the moment to start inter-frequency scanning.The problem of joint optimization of Inter-frequency Measurement Power(IMP)and Traffic Offloading Opportunity Utilization(TOOU)is then transformed into a problem of approaching the coverage of a wide-band small cell with the detection region.By introducing recall and precision,the above problem can be further transformed into a threedimensional space search problem,which is solved by an algorithm based on particle swarm optimization.The results,which are the optimal thresholds of the detection region,can then be used to determine which moment is the best to start inter-frequency measurement.Simulation demonstrates that the proposed scheme can reduce more than 37%IMP in low mobility states under the promise of over 90%TOOU.Second,a computation offloading mechanism on the access side of IIoT is proposed,in view of problems of data freshness,device fairness,and the deep coupling among communication,energy,and computing resources.From a real-world implementation perspective,it facilitates scheduling in the presence of partial and outdated Network State Information(NSI).In specific,a system utility maximization problem that consists of throughput,fairness,and data age is presented.Asymptotic optimization is applied to decouple the system utility maximization problem into three independent sub-problems:fresh data collection problem,stale data discard problem,and data offloading problem.Leveraging convex optimization techniques and Lambert W function,closed and semi-closed solutions to the above problems can be obtained.Moreover,the optimality loss brought by asymptotic optimization is analyzed,especially under partial and outdated NSI.Extensive simulations show that the proposed approach can improve 33%of system throughput while ensuring data freshness and device fairness.Third,an online resource scheduling scheme for the access side of heterogeneous IIoT is proposed.The influence of non-negligible NSI costs on system scheduling is analyzed considering real network requirements.Specifically,a system utility maximization problem comprising throughput and fairness is presented,subject to different energy constraints of devices.A perturbed Lyapunov technique is employed to decouple the problem into slot-based deterministic subproblems of data collection,energy utilization,and data offloading,avoiding the demand for non-casual NSI.The optimal scheduling decisions of the sub-problems are derived by using convex optimization techniques.Its performance under non-negligible NSI feedback cost is also analyzed.Simulations validate that,compared with benchmarks,the proposed scheme can increase at least 15%of throughput and tackle the doubly near-far problem as well as the unfairness caused by device heterogeneity.Fourth,a repair mechanism for damaged systems on the network side of IIoT is proposed.In addition,an algorithm that can reduce computational complexity in the case of repairing heavily damaged networks is presented.In specific,a system cost minimization problem comprising both repair costs and computational demands is formulated.Based on Benders decomposition,the mixed-integer linear programming problem is transformed into a main problem for link recovery selection and a sub-problem for computation migration.The optimal solutions can be obtained through the iteration of solving the two problems.Moreover,a local branching method is introduced to accelerate the convergence of Benders decomposition.Simulation results demonstrate that the proposed approach achieves better performance in multiple topology scenarios,which results from the consideration of network dynamics.For random networks,nearly 50%of system costs can be reduced compared with the giant-component-based approach.