Research On The Transmission Mechanism Of Cloud-Edge Services With High Service Quality Assurance

Emerging services such as the Internet of Things,Big Data,Augmented Reality,and Virtual Reality have sprung up like mushrooms after a rain,and the business volume in the communication network has also shown explosive growth,which puts forward higher requirements on the delay,throughput,energy consumption and safety guarantee of the communication backbone network.At present,the emergence of the Cloud-Edge collaborative computing meets the low-latency and high computing requirements of network services.Moreover,with its fine-grained spectrum division,flexible resource scheduling method,and efficient resource utilization,the Elastic Optical Network(EON)has become the substrate network with the most development potential for the Cloud-Edge collaborative network by virtue of fine-grained spectrum division,flexible resource scheduling,and efficient resource utilization.However,the unbalanced traffic demand of large-scale Cloud-Edge services will inevitably cause a large amount of spectrum fragmentation,which will lead to additional energy consumption.Traditional energy-saving strategy feature extraction is not easy,and lacks intelligent decisionmaking.In addition,the security problems of the Cloud-Edge collaborative network itself also need to be solved urgently.Traditional security assurance technologies rely on layer-by-layer isolation and have not actively improved the security of the network from the level of system consensus.Therefore,this paper focuses on the two aspects of Cloud-Edge collaborative network traffic grooming and security assurance to achieve the high energy efficiency and resource change security of the Cloud-Edge collaborative network,so as to providing high-quality Cloud-Edge services.The specific contents are as follows.1.To improve the self-adaptive grooming service capability and energy efficiency of the Cloud-Edge collaborative elastic optical network,this paper proposes an Energyefficient Deep Reinforced Traffic Grooming(EDTG)algorithm,which adaptively learns grooming strategies according to the traffic demand of the service to reduce the energy consumption of the Cloud-Edge collaborative network.The main contributions are as follows: EDTG merges the Actor module and the Critic module into one network,reducing network parameters and complexity.It updates the shared network by calculating the total loss and starts multiple processes to speed up training.Simultaneously,different from the existing methods of manually extracting network features,EDTG converts the network modalities and business modalities into colored pixel images and uses Mobilenet V3 to extract the features to ensure the integrity of the information.Besides,through in-depth analysis of various traffic grooming situations and repeated experiments,this paper establishes a reward and punishment mechanism with 12 categories to optimize the grooming strategy of the EDTG continuously.The simulation results demonstrate that compared with two well-performing baseline algorithms,Deep Reinforcement Learning(DRL)and State-aware Modification Grooming Algorithm(SGA),EDTG reduces energy consumption by 13% and 8%respectively,effectively reducing network operating costs.2.To enhance the security and reliability of the Cloud-Edge collaborative elastic optical network when resources are changed,this paper proposes a Cloud-Edge Domain-specific consensus mechanism based on Practical Byzantine Fault Tolerance(CED-PBFT),which divides the Cloud-Edge collaborative network into different domains.The collaborative consensus of intra-domain and inter-domain nodes ensures that every resource change is safe and reliable across the entire network.The main contributions are as follows: CED-PBFT establishes a mutual trust model to select the master node mechanism.A mutual trust model is established based on the consensus history between nodes.The node with a higher public trust value is selected as the master node,which uses node computing resources in a balanced manner and reduces the probability of the master node doing evil.In addition,CED-PBFT divides the Cloud-Edge network into different domains and adopts an intra-domain inter-domain collaborative consensus strategy instead of all nodes in the entire network participating in consensus at the same time,which significantly reduces the verification complexity of the consensus algorithm,and improves the consensus efficiency and the scalability of the network.