Automated silent eNode B detection
Identifying hidden problems for a Japanese operator
Silent eNB & KPI degradation detection using machine learning algorithms for improved efficiency and network quality.
Silent eNB (or Sleeping Cell) is a well-known problem that affects network equipment from all vendors.
- The base station is unable to carry traffic resulting in a service degradation but appears to be functioning normally > No Alarm is triggered!
- Identifying the issue and detecting the root cause was a very manual, time-consuming process for the customer: taking up to 24 hours after the event and only yielding an 80% success rate
- Use machine learning algorithms to identify and detect the root cause
- 6 month Proof of Concept in Tokyo area to prove Nokia solution and fine-tune algorithms
- Solution created following lean software development principles
- Designed to meet traditional Japanese quality standards: 100% accurate detection and reporting in Near-Real-Time (less than 30 minutes)
Outcome & benefits:
- 100% success rate for detection of Silent eNodeB, within 30 minutes
- In addition to identification of silent eNB cases & several hardware faults were also detected
- Automatic root cause analysis generated to help resolve issues
- No end user complaints received during the PoC period
- Proved to be much more effective & efficient than existing customer process
- Results acknowledged by customer and PO received for full nationwide service covering Nokia, Ericsson & Samsung