April 24, 2026

Enhancements to Predictive Analytics for Optical Assurance

The Predictive Analytics feature for Optical Assurance has been enhanced to forecast and identify outliers based on Bit Error Rate (BER), Forward Error Correction (FEC), and Cyclic Redundancy Check (CRC) metrics.

The ML models use a custom probe added in the DC-Probes analytics pack to collect data from the Data Center Director. You can install the DC-Probes pack from DC Hub page. The models use this data to predict future metrics and recognize outliers that might occur in a 15-day period for the following parameters:

• FEC Corrected Codeword Rate

• FEC Uncorrected Codeword Rate

• Pre-FEC BER

• RX CRC Errors Rate

• TX CRC Errors Rate

• FEC Histogram

Note:

Predictive Analytics for Optical Assurance based on BER, FEC, and CRC metrics requires Data Center Director 6.1 and devices running the following Junos OS Evolved releases: 23.4X100D31, 25.2X100D10, and 25.4R1.

[See Predictions Overview.]

Reliability Metrics and Bubble Chart View for Predictions

The Predictive Analytics feature has been enhanced to include confidence metrics for an outlier and display predicted outliers in the form of a bubble chart. It has also been enhanced to display a prediction band in the parameter graphs on the Device Details page.

The confidence level represents the reliability of the predictions in percentage.

The bubble chart maps the predicted time the issue might occur and the confidence level for an outlier. Each bubble represents an outlier on a network device. The color of the bubble indicates the severity of the outlier that is determined based on whether the forecasted value exceeds the system-defined thresholds. The size of the bubble indicates the number of services or clients impacted by the outlier. You can use filters to modify the bubble chart as per your requirements.

The prediction band in the graphs for each parameter indicates the range of values within which a future observation is likely to fall based on the confidence level of the forecasting method. A narrow prediction band indicates greater precision and confidence in the forecast. A wide prediction band indicates uncertainty and variability in the forecast.

The confidence level metrics, the bubble chart, and the prediction bands help you identify and prioritize critical issues in the network. This enables you to take corrective actions in advance and prevent network disruption.

[See Predictions Overview.]