摘要：In this paper, an algorithm for predicting accessibility performance on
an LTE/SAE network based on relevant historical key performance indicator
(KPI) data is proposed. Since there are three KPIs related to accessibility, each
representing different segments, a method to map these three KPI values onto the
status of accessibility performance is proposed. The network conditions are
categorized as high, acceptable or low for each time interval of observation. The
first state shows that the system is running optimally, while the second state
shows that the system has deteriorated and needs full attention, and the third
state indicates that the system has gone into degraded conditions that cannot be
tolerated. After the state sequence has been obtained, a transition probability
matrix can be derived, which can be used to predict future conditions using a
DTMC model. The results obtained are system predictions in terms of
probability values for each state for a specific future time. These prediction
values are required for proactive health monitoring and fault management.
Accessibility degradation prediction is then conducted by using measurement
data derived from an eNodeB in the LTE network for a period of one month.
其他摘要：In this paper, an algorithm for predicting accessibility performance on an LTE/SAE network based on relevant historical key performance indicator (KPI) data is proposed. Since there are three KPIs related to accessibility, each representing different segments, a method to map these three KPI values onto the status of accessibility performance is proposed. The network conditions are categorized as high , acceptable or low for each time interval of observation. The first state shows that the system is running optimally, while the second state shows that the system has deteriorated and needs full attention, and the third state indicates that the system has gone into degraded conditions that cannot be tolerated. After the state sequence has been obtained, a transition probability matrix can be derived, which can be used to predict future conditions using a DTMC model. The results obtained are system predictions in terms of probability values for each state for a specific future time. These prediction values are required for proactive health monitoring and fault management. Accessibility degradation prediction is then conducted by using measurement data derived from an eNodeB in the LTE network for a period of one month.
