A method to minimize the maintenance costs of a broadband access network based on optimal steady-state probability mixture distribution of proactive and reactive maintenance

Abstract

• Minimizing the maintenance cost of access lines in a broadband multimedia network. • A large set of line failure data collected over a period of 10 years is considered. • Line failure occurrences can be described by a probability mixture distribution. • Optimal maintenance; reactive and proactive approaches included in a balanced way. • Optimal share of proactive maintenance calculated by a discrete-time Markov chain. Paper presents an analysis of the long-term maintenance of access lines in a real broadband network. The dynamics of the maintenance process were recorded by using fault management systems over a ten-year period. The data collected belong to three groups: data that represent the stochastic nature of the occurrence of failures on lines, data related to the type and quality of detector used for remote detection of faults and degradation based on Quality of Service (QoS) measures, and data on user habits that connect QoS and Quality of Experience (QoE). We devised our model proposal that enables the calculation of the optimal amount of proactive maintenance, taking into consideration the characteristics of the network, characteristics of the detector system and QoE. The intention was to facilitate decision making about optimal ratio of reactive and proactive maintenance, which reduces costs. Ratios are calculated as a stationary probability vector of a Markov chain, which is finally used to calculate the optimal maintenance cost. Maintenance cost reductions can vary, depending on the different variable settings of the model, but in the case covered in the paper, with 8.33% of proactive repairs introduced, maintenance costs were reduced by 3.71%.