Research On Fault Diagnosis And Performance Optimization Algorithm Of Communication Backbone Network

The communication backbone network is a critical infrastructure for providing high-availability services to human society and shares the characteristics of providing information exchange with the communication network,but with a larger scale and a higher level of importance.In the era of universal communication and the Internet,the population is increasingly demanding the quality of communication networks.However,network systems are constantly facing a range of challenges due to misuse,environmental changes,configuration errors,etc.,often falling link or node components.Also,the share of large-scale failure scenarios that affect multiple network elements simultaneously in a geographically correlated manner,such as disasters and malicious attacks,is increasing significantly in recent years.The various faults in communication networks and their physical characteristics,such as hidden root causes,propagation,and regional impact,can not only lead to deterioration in network quality,but can also cause significant financial losses.It is a major challenge for network managers to quickly diagnose the root cause of the fault after the occurrence and to take various measures to mitigate the negative impact on network performance.To this end,this paper focuses on fault diagnosis algorithms,fault propagation control strategies,as well as their vulnerability and resilience in the face of regional failures,for communication networks.These tasks are of great importance to maintaining and optimizing network performance.The advent of the information age makes it easier to access information about network failures,but the current problem is that the data contains a lot of useless information.How to efficiently pre-process the data and diagnose the root cause of faults is still a challenge.Given the causal relationship within the data,a Bayesian Network(BN)based fault diagnosis algorithm is proposed in the context of massive data.A silent gap-based data segmentation method is adopted to handle the data asynchrony during pre-processing.More importantly,the BN for fault diagnosis is constructed with the help of temporality,and the root cause of the fault is reasoned on this basis.Simulation results on real datasets show that the data segmentation algorithm reduces the runtime while improving the sequence segment availability and that the proposed fault diagnosis mechanism allows the identification of the fault source with higher accuracy and faster running speed.In particular,the algorithm module is embedded in an efficient processing framework to maximize its effectiveness.Besides,an effective propagation control strategy is required since a single fault always propagates and spreads rapidly through the network,even causing localized downtime.The fault propagation is controlled by intentionally recovering the low overload failure edges,which enhances the transferability of information in the network while reducing the negative impact.In addition,previous work on recovery-based propagation suppression only considers failed edges caused by overload,while the fault location task can identify non-overload-induced failure edges.Therefore,given the availability of fault location results,the control of fault cascade is improved by recovering different types of failed edges.Different recovery strategies are compared in simulation experiments to demonstrate the effectiveness and feasibility of the proposed algorithm.Increasingly frequent regional failures are devastating to the reliability of communication networks and high transmission links.A key aspect of research to mitigate the impact of disasters on network performance is how to measure the vulnerability of communication networks in disaster scenarios.The premise of vulnerability assessment is to establish a reasonable and fine-grained failure model,due to the special evolutionary mode of the communication network under regional failures.A new region failure probabilistic model is proposed comprehensively considering the influence of two uncertainties,distance and link length,on the failure probability.In addition,two approximate assessment algorithms,namely the reduction of the candidate disk number based on geometric transformation and grid partition,are applied to reduce the complexity of the vulnerability analysis.Simulation results on real communication network exhibit that the proposed probabilistic model provides a more realistic description of network behavior and reflects,to some extent,the resilience of the network itself.Because of the severe and extensive impact caused by regional failures,mitigation of their impact needs to be addressed in the network design phase.In terms of regional fault resilience optimization,the communication network is designed to be highly resilient,which means its preparedness for disasters is improved.Specifically,grounded in the essential characteristics of regional failures,geographical diversity is taken as an indicator of the network’s resilience in the event of disasters.Then,we design communication networks with high geodiversity by shielding network components,one of the most attractive advantages of this approach is its implementability,rather than just being a theoretical technique.Finally,the minimum cost shielding problem with ideal geodiversity is constructed as an Integer Nonlinear Programming(INLP)model.Considering the NP-hardness of the problem,a heuristic algorithm is adopted to obtain an approximate optimal solution at a low computational cost.Simulation experiments confirm the feasibility of the disaster resilience optimization model and the usability of the geodiversity concept under regional failures.In summary,this paper investigates the fault diagnosis problem and related network performance optimization problems in communication networks,proposes new solutions,and actively performs simulation experiments to verify the scientificity,rationality,and excellence of the proposed algorithms and research ideas.