Building automated closed-loop operations in your optical network
In our last blog post, we outlined the key ingredients needed for automated service provisioning in optical networks. We also explained how open API interfaces can lower network TCO by simplifying existing business processes and making them more intent-driven.
Now that we know what we need for automated service provisioning, let’s concentrate on building automated closed-loop operations.
About closed-loop control systems
Closed-loop control systems have been implemented for years in various common applications. These systems automatically and precisely tune electronics to maintain a desired state without any external human intervention. Any closed-loop application requires intelligent KPI sensors as inputs, which the control system uses and reacts to.
A common example is cruise control in a car. The car’s sensors provide speed as the KPI that is monitored, and the control system provides the necessary actions to keep the speed at the desired level.
In optical dense wavelength division multiplexing (DWDM) networks, one of the main applications for control loops is per-channel optical power along with amplifier gain. The control system provides the actions to maintain wavelengths at a desired power level while also ensuring that the optical amplifier achieves its desired state to compensate for link span losses in the network. Gain and power become the measured KPIs required by the control system for sensing and processing.
Steps in automated closed-loop operations
To achieve performance-sensing intelligence in your optical network, you need to:
- Monitor the appropriate set of KPIs
- Ensure the collected KPI data is reliable
- Perform deep analysis, trending and prediction
- Use appropriate optimization-tuning techniques
- Employ an automated workflow framework to automate this process.
The results are automated closed-loop operations.
Why bother? If you can create targeted, automated closed-loop operations, you can focus on specific use cases that are of key interest to your business. By leveraging various KPIs as the sensing data used in the feedback of the closed-loop automation workflow, you can proactively address potential issues that may appear in your optical network over time. You can also use automated closed-loop operations to streamline your network life cycle and rapidly build new capacity where it’s most needed.
Key requirements for automated closed-loop operations
As shown in Figure 1, key requirements for automated closed-loop operations that offer high value for network operators are:
- Availability of KPIs from the optical network
- Trending and multivariate analysis data science expertise
- Optimization capabilities
- Workflow management capabilities as part of the automation framework.
Figure 1. Requirements for automated closed-loop operations
Having a wide range of measured data (KPIs) from the optical network elements at various geo-locations provides valuable insights into what is happening in different segments of the network. As the measured data is collected, it is important to index this data over time so that the collection can be stored in a database for the purpose of trending changes over different time periods.
This time-series database is one of the foundations for automated closed-loop operations. While optical KPIs are important, often other sources of information such as external factors— environmental or contextual information—can also be analyzed and used for correlation.
At Nokia, the WaveSuite Health and Analytics (WS-HA) application is the centralized repository for this data capture. Nokia WS-HA is used to visualize and analyze the information in the network so it can self-learn and predict events before they begin to impact the SLAs of end- customers.
Trending and multivariate analysis
As the time-series database is built up with measured KPIs, the collection of data can be used to discover trends in optical events or impairments. You can capture and classify the events/impairments in your network, then execute corrective actions.
Because a series of different KPI parameters are captured, you can perform trending for each. You can also perform multivariate-based correlation analysis to learn more about the statistical significance of one parameter compared to the others. In addition, the trending and multivariate analysis provides a mechanism to set threshold limits for individual KPI parameters or a correlation of multiple KPI parameters to gauge optical network performance levels.
Automation framework with workflow management
Developing a workflow that includes communication between applications enables the triggering of digitized procedures to react to the measured data from your optical network—so you can achieve your desired business outcome. REST notifications can be used to provide event messages, which establishes this necessary communication.
In a workflow use case that requires the outcome of optimized performance, for example, the event notification can request WaveSuite Optimizer to recommend an optimization method that best addresses the link degradation. The workflow processes the recommendations and invokes a new process that allows a modified set of parameters to be provisioned in the network to recover the lost margin.
Nokia WaveSuite Optimizer, WS-HA and WaveSuite Network Operation Centre (NOC) can all generate notifications required by workflow systems, enabling closed-loop operations.
WS-HA also enables other closed-loop workflow processes based on specific KPI trending behaviors or on trending correlated multivariate KPIs. For example, KPI threshold trigger notifications, which target predictive capacity upgrades on segments of the optical network, can stimulate planning for new optical wavelengths to service the required capacity growth.
Optical networks undergo a variety of changes that degrade performance over time. Stress on the outside plant fiber, aging equipment or extreme environmental factors typically translate into an increase in fiber attenuation or a decrease in signal quality: both impact WDM systems. The result is degradation of the available margin on a given optical segment, which can affect end-customer services. Optimization techniques are necessary to compensate for these events in your network and to regain the margins.
Nokia’s coherent next-generation digital signal processors (DSPs) and high-performance reconfigurable add-drop multiplexing (ROADM) technology provide you with various forms of optimization:
- Channel optimization: A variable baud rate transceiver is changed to modify its required optical signal to noise ratio (OSNR) tolerance. (For more details, see our blog post Automated power optimization: autofocus for optical networks.)
- Link optimization: Optical parameters such as gain, tilt/ripple and channel power are changed to modify link degradations caused by a change in link loss.
- Defragmentation and adaptive wavelength allocation: Channels are spectrally positioned to make better use of the spectrum and improve the overall OSNR performance.
The Nokia WaveSuite Optimizer application supports the previous optimization methods and also supports open interfaces and notification advertisements. As a result, Optimizer can be easily integrated into closed-loop workflows.
Now that we know the key requirements for automated closed-loop operations, future blog posts will discuss specific automation use cases that can help you better diagnose and troubleshoot network problems—to deliver on existing SLA commitments and protect revenue streams. Stay tuned.