Classification,Quantification,and Prevention Of Human Errors In Communication Network Operation And Maintenance

In the sector of information and communication technology,network operation and maintenance are necessary to ensure the normal operation of hardware and software,maintain a high level of availability and reliability of the network system,and reduce the cost of failure.Statistics have shown that human error is one of the most important causes of network outages.With the increments of popularity and complexity of network systems,the impact of human error becomes increasingly significant and receives much attention.However,traditional research on human error in the human factor and safety community mostly focused on typical safety-critical systems such as nuclear industry,aerospace,and medicine,while little has investigated human errors in the information and communication technology sector.To fill in this gap,the present study conducts systematic research on human error in network operation and maintenance and develops a complete framework for human error identification,prediction,and control.There are three main researches reported in this dissertation.(1)A task analysis of typical network operation and maintenance tasks is firstly conducted which identifies the human-computer interaction characteristics,and the generic task types are categorized based on that.Moreover,potential human errors in these generic task types are identified through domain expert judgment,and different error modes are classified based on the cognitive failures and behavioral manifestation of errors.(2)For identified error modes,the likelihood is obtained through domain expert judgment,and the judgment values are mathematically aggregated through Bayesian approaches.(3)Risk matrices are developed to evaluate the risk of human error based on the likelihood of occurrence and the severity of consequence.Finally,technical and managerial measures for human error control are proposed as different system barriers.The main results obtained include:(1)A generic task type structure for network operation and maintenance,which consists of three attributes including operation,system object,and system function,with a total of 6 operations,14 system objects,25 system functions and 66 generic task types through the combination of the three attributes;in addition,a three-level erroneous behavior taxonomy for network operation and maintenance,including 10 Level 1 error modes(failure of macro-cognitive functions),40 Level 2 error modes(behaviorally observable errors)and 80 instanced error modes(behaviorally observable errors in specific tasks);(2)basic human error probability for the 40 Level 2 error modes;(3)human error risk matrices for two types of adverse events(service quality events and data obliteration events),and technical and managerial measures for human error control as four different barriers(avoiding human errors,reducing human errors,error tolerance,and accident response).Overall,the present study proposes a systematic framework for human error identification,prediction,and control for network operation and maintenance.Several key technical details,such as three different attributes of generic task types,a human error classification paradigm that integrates cognitive and behavioral perspectives,and the human error probability of Level 2 error modes,are meaningful for research on human error.In addition,this framework can help network managers systematically analyze operation and maintenance tasks,identify potential human errors,assess human error risks,and develop error control strategies,to reduce human errors in such tasks and improve system reliability.The proposed framework has been successfully applied in a technology company to guide the design and management of network operation and maintenance tasks.