Project Overview
The initial submission information can be added to this section as a starting point. Please note any use case tuning and updates since the submission timeline. Please ensure the text is concise and each topic remains within a guideline of 400 words for each section.
The champions will understand the importance of abiding to industry standards and their inter-relationship. Several questions will eventually be answered, including:
How can the data of a particular cell be quickly and dynamically diagnosed as abnormal at a specific time? Which device manufacturer is more likely to show data abnormity in their devices? When will the data from the device manufacturers change dramatically? How can the cells that need capacity expansion be identified using index data?
These problems will be solved by establishing an accurate and applicable model with AI algorithms. The model could assist operators in achieving more efficient operation and maintenance.
At present, AI technology has been applied in various industries, but there are few commercial applications of AI technology in the field of wireless network O&M and optimization which are mostly labor-intensive process. A large number of manual work is needed to guarantee the network quality. Despite the high labor costs, there is still a great lag for network problem detection, even major accidents cannot be found out before it happens.
First of all, the judgment of abnormity in historical data mainly depends on the criteria set by experienced personnel. Such criteria often remain unchanged all the year round and cannot be applied to complex and changeable cells around the country. More importantly, it is difficult for operations staff to find abnormal historical data in time, which may lead to major data accidents.
Secondly, the same index might be measured and calculated in different ways by different equipment manufacturers, which makes it impossible to compare the data directly. In addition, the equipment manufacturers will upgrade and maintain the devices periodically, which may also cause a change in the index. However, such changes will not be reported to the operators in time, which will also makes it difficult for the operation and maintenance.
Finally, whether to expand the cell capacity is determined by the prefixed criteria for multiple indicators, combining with the verification of local operations staff. Thus, the capacity expansion judgment exists problems such as single standard and relatively long procedure.
In this project, three AI algorithms, anomaly diagnosis, trend prediction and capacity expansion prediction, are established to assist the planning, maintenance and optimization of wireless network so as to improve network operation efficiency and reduce costs.
Project Vision Statement
AI-based intelligent detection and prediction solution. Improve quality before it deteriorates, manage traffic before congestion occurs!
What business challenges are you addressing?
If there is more than 1 user story portrayed please copy the template below or ensure the user story answers the following: Who is the user?; What needs to be achieved?; For what purpose?; How it will be done?; How success is measured.
As an __ operation and maintenance personnel___
I need to _ detect and predict the possible network failures __
so that I can __ improve the efficiency of network operation and maintenance __
To do this I need to __capture and understand the patterns of historical network data__
I know that I am successful when _ network failures can be predicted and located from data __
What makes this a significant problem to be solved?
At present, AI technology has been applied in various industries, but there are few commercial applications of AI technology in the field of wireless network O&M and optimization which are mostly labor-intensive process. A large number of manual work is needed to guarantee the network quality. Despite the high labor costs, there is still a great lag for network problem detection, even major accidents cannot be found out before it happens.
First of all, the judgment of abnormity in historical data mainly depends on the criteria set by experienced personnel. Such criteria often remain unchanged all the year round and cannot be applied to complex and changeable cells around the country. More importantly, it is difficult for operations staff to find abnormal historical data in time, which may lead to major data accidents.
Secondly, the same index might be measured and calculated in different ways by different equipment manufacturers, which makes it impossible to compare the data directly. In addition, the equipment manufacturers will upgrade and maintain the devices periodically, which may also cause a change in the index. However, such changes will not be reported to the operators in time, which will also makes it difficult for the operation and maintenance.
Finally, whether to expand the cell capacity is determined by the prefixed criteria for multiple indicators, combining with the verification of local operations staff. Thus, the capacity expansion judgment exists problems such as single standard and relatively long procedure.
What will change about the way your champions do business if this problem is solved?
The champions will understand the importance of abiding to industry standards and their inter-relationship. Several questions will eventually be answered, including:
How can the data of a particular cell be quickly and dynamically diagnosed as abnormal at a specific time? Which device manufacturer is more likely to show data abnormity in their devices? When will the data from the device manufacturers change dramatically? How can the cells that need capacity expansion be identified using index data?
These problems will be solved by establishing an accurate and applicable model with AI algorithms. The model could assist operators in achieving more efficient operation and maintenance.
Resourcing of the Project
This is a mandatory section for each Catalyst Project. The goal of this section is to clearly identify all the resources and skills required by the project to successfully deliver the demonstration and deliverables in Section 2.
Please identify the companies and specific people committed to participate in this work and the role to be filled by each participant. (Feel free to adapt this table to suit your Project's needs.)
Note this section will also be used as input for Best new Catalyst in show award.
Member Project Resources and Roles
Role | Name | Company | Key Contributions (List these concisely in no more than 3 bullet points per row) |
|---|---|---|---|
Project Co-Leader | Wang Bing | China Telecom |
|
Marketing Lead | Lu Yan | AsiaInfo | Business negotiation Marketing Product copywriting |
Champion 1 | Qian Bing | China Telecom Beijing Research Institute |
Scheme writing and modification |
Participant 1 | Zhiying Yin | AsiaInfo | Exception algorithm development Platform development |
Participant 2 | Songyan Gao | AsiaInfo | Product design Database development |
Participant 3 | Guan Hao | BOCO Inter-Telecom | Capacity prediction algorithm development |
Participant 4 | Tian Xin | BOCO Inter-Telecom | Marketing |
Participant 5 | Yimin Yuan | HUAWEI | Marketing Scheme writing and modification |
Participant 7 | Zekun Dong | HUAWEI | Short cycle capacity prediction algorithm development Long cycle capacity prediction algorithm development |
Participant 8 | Giles | HUAWEI | Scheme writing and modification |
Participant 9 | Wang Dan | ZTE Corporation | Exception algorithm development Platform development |
Participant 10 | Zhongzhi Lin | ZTE Corporation | Index consistency test algorithm development Platform development |
Participant 11 | Chunyuan Zhang | ZTE Corporation | Marketing Scheme writing and modification |
TM Forum Mentor | Shengfan Hou | TM Forum | N/A |
Measuring business impact results
Business Impact / Business Value of the Catalyst solution
This Please complete each row with a business impact statement and provide a description for each statement in a maximum of 150 words. Please provide tangible expected results such as credible cost savings or additional revenue a champion might observe by implementing the Catalyst solution.
Note section 4.1 & 4.2 will also be used as input for Outstanding Catalyst – business impact & Best new Catalyst in show award.
| Business Impact | Description/ Proof points |
|---|---|
80% MTTR reduced 20% NOC efficiency improved | 80% Mean Time to repair is reduced, 20% NOC efficiency is improved due to detect problem and locate the root cause in advance.
|
Clean 15-30% of dirty data | Through data quality control technology and index consistency checking algorithm, 15-30% more unqualified data were cleaned compared with the traditional pretreatment method. |
Abnormal data discovery time reduced by 96% | Through the three-step anomaly detection method of single dimension detection->expert rules->multidimension detection, the cell performance anomaly detection without fixed threshold is realized. Compared with the traditional manual method to detect abnormal data, the time is reduced by 95%. |
The identification time of the expanded cell was reduced by 80% | Through the long and short cycle capacity prediction algorithm developed by the project, the identification time of the expanded area is shortened by nearly 80%. |
Commercial Impact for the champion or broader market within 12 months.
Please answer this section with a Maximum of 150 words. In this section the catalyst team is asked to provide information on how likely is that the concept will become commercially viable in the next 12 months.
Commercial application planning
- Manage and monitor data quality of 1 million equipment nationwide;
- Applying to 3 million Telecom communities and 200 million wireless network users nationwide to improve the optimization and O&M efficiency.
Measuring Innovation
Business technology solutions innovations.
Please complete each row with an Innovation statement and provide a description for each statement in a maximum of 150 words. Please provide evidence of the Project's business / technology originality.
Note section 5.1 & 5.2 will also be used as input for Outstanding Catalyst – innovationt & Best new Catalyst in show award.
Innovation | Description/ Proof points |
|---|---|
Automatic identification of inconsistent data from different OMC vendors | Through index consistency checking algorithm, achieving automatic identification of inconsistent data from different OMC vendors, this is difficult to achieve in o&m management in the past |
The accuracy of the anomaly diagnosis algorithm was improved to 89% | Through data quality control technology and index consistency checking algorithm, The accuracy of the anomaly diagnosis algorithm was improved to 89%. |
Short period traffic prediction effectiveness to 97.12% | Through data quality control technology, index consistency checking algorithm, and AI algorithm removing abnormal data, short period traffic prediction effectiveness to 97.12%. |
Long period traffic prediction effectiveness to 81% | Through AI algorithm removing abnormal data, cell grouping modeling based on spatiotemporal characteristics, autoML search for the optimal combination of cell groups, Non-time-sequence prediction algorithm reduce long period error accumulation and improve prediction accuracy. Long period traffic prediction effectiveness to 81%. |
Potential innovation impact.
Please provide information on the benefits and potential impact (commercial, societal...) of the business or technology innovations developed during the Catalyst phase. Please answer this section with a maximum of 150 words.
5G deployment is accelerating. 5G site planning will be more accuracy with traffic prediction.
TM Forum Assets
Usage of TM Forum Assets
Please select Frameworks / Open API / other asset usage. Please limit the detail for each item to 150 words per asset.
Note section 6.1 will infomation will be used as input for Outstanding Use of TM Forum Assets award.
| ASSET | How did you use it? | Describe or quantify the benefit / what did you learn? |
|---|---|---|
| 1.OPENAPI | ||
| 2. | ||
| 3. | ||
| 4. | ||
| 5. |
Contribution to TM Forum Assets
Contributions to TM Forum Assets can be made up to 6 weeks after the project end date. Each contribution must be captured in Jira aligned with a TM Forum workstream before being entered in this table for review. For further information on how to make a contribution visit our Catalyst Contributions help page.
Note section 6.2 will infomation will be used as input for Outstanding Contribution of TM Forum assets award.
| CONTRIBUTION | ||||
| ODA / ODF Assets | Best Practices | OPEN APIs | Use Cases | Other - i.e Report, Catalyst Whitepapars |
Please inster the Jira contribution link | Please inster the Jira contribution link | Please inster the Jira contribution link | Please inster the Jira controbition link | Please inster the Jira controbition link / plugged in document. |