The Avanseus CAN remedies the challenges, inefficiencies, and instability faced by a growing network by predicting potential failure of network events and recommending preventive measures. It helps to detect failure that cannot be identified by existing network monitoring systems and uses advanced machine learning to continuously extract fresh information for efficient network operations. Avanseus CAN is able to analyze syslog messages, trouble ticket information, alarm information, network configuration data, inventory data, and customer complaints as well as external data such as weather conditions. The solution analyzes this data, recommends action, improves network availability, and reduces the need to spend time on internal root cause analysis.
- Prediction of network incidents in advance
- The Avanseus CAN (Cognitive Assistant for Networks) platform automatically predicts future faults in the network 7 to 30 days in advance (configurable) and recommends the root cause of the failure based on historical fault patterns and aid NOC and Field Teams to effectively manage these failures by proactively performing preventive actions. Prediction can be done with the following granularity:
• Equipment Identifier (Equipment for which failure is predicted)
• Site (Physical location of the equipment)
• Reason for impending failure (fault type or fault group such as cell down, node down, VSWR threshold crossed etc.)
• Perceived priority of the impending failure (such as Priority 1, Priority 2 and so on)
• Probability of occurrence
• Other information to enable preventive action including root cause recommendation, clustered events related to same root cause, correlated events
- Predicting KPI degradation
- The platform automatically predicts network performance KPIs and identify the potential threshold violation cases. The process flow is suggested to be as follows:
• Platform collects PM counters / KPI data from a KPI data lake or source systems in a “near real time” manner, may be at a frequency of 15 minutes or 1 hour depending on the capability of the provider system.
• It also collects the “thresholds” associated with each PM KPI as defined in the provider system.
• Platform is able to predict possible KPI violation (in comparison to defined “threshold”) in next ‘X’ hours. The frequency of prediction will be mutually agreed before project implementation based on the way it is envisioned to be operationalized.
• Platform will send the information related to possible KPI violation and the probable Root cause to the relevant service orchestration system for corrective action.
The platform can predict KPIs for any technology or domain.
- Health index monitoring and predictions of VNFs
- While 5G and Telco cloud will have large scale implementation of NFVs running on commodity hardware, the downsides are virtualised network functions (VNFs) and their host commodity servers are more failure prone than dedicated hardware, and virtualization introduces more layering and less visibility into lower layer faults. One question critical to the success of NFV systems is whether it can provide availability similar to that of traditional carrier-grade systems, with up to five 9s (99.999% of uptime). This requires AI to provide relevant insights and suggested actions in a fast and reliable way.
Avanseus CAN is able to address these challenges by providing the functionalities such as, performance predictions of Virtual Network Functions(VNFs), Continuous monitoring of services (health index), prediction of rare faults, very fast prediction engine to cater to manifold increase in services, integrated with orchestration engines for automated capacity provisioning.
