The problem many communications service providers face in leveraging data for network automation ties directly to their reliance on siloed architectures. Find out more in this preview of our upcoming report Network automation using machine learning and AI.
30 Mar 2020
Sharing data shouldn't be so hard
TM Forum’s new report Network automation using machine learning and AI shows the importance of applying automation to processes across a communications service provider’s (CSP’s) entire organization, from the network to business units. Many of the processes will require artificial intelligence (AI) to achieve closed-loop automation, but whether AI-enabled or not, automation relies on data for intelligent decision-making to adjust parameters, alter configurations and understand users’ behavior.
The problem many CSPs face in leveraging data ties directly to their reliance on siloed architectures. Silos inherently keep data locked within. As it turns out, architectural silos breed organizational silos, and sometimes it is less a technical barrier to sharing data between groups as it is a political or territorial one. Group leaders are sometimes reluctant to share data with other groups because they fear it will be abused, especially customer data. This is untenable.
Lack of standards
In a survey conducted for our report, we asked CSPs around the globe why sharing data is so difficult. The top challenges are lack of defined and understood standard interfaces, identified by a full 81% of CSPs as a moderate or significant challenge, and not having a clear understanding about which data can and cannot be shared, identified by 77%.
Data repositories may reside with third-party data warehousing providers, cloud-based analytics providers or internal analytics systems. Although TM Forum and others are working on open application program interfaces (APIs), there is not yet a standard for interfacing with repositories that may format data differently.
Which data is OK to share?
Not having a clear understanding of which data can be used or shared outside the organization with partners is a significant problem, which is also closely related to the No. 3 and No. 4 challenges of regulatory compliance and traceability. Not to be confused with explainability, which is understanding why AI systems come to the conclusions they do, traceability refers to CSPs’ concern about what becomes of data once it’s shared with authorized partners. Consumers can opt out of having their usage and behavior shared, but for those who do not, CSPs can share anonymized location, browsing and application usage information with marketing partners to develop targeted advertising.
Operators need be able to verify that the data is not being shared with any non-authorized parties or further analyzed to try to associate the data with specific users. Partners must adhere to the rules under which the CSP operates, because the CSP is responsible.
Addressing cultural issues
While uncertainty about internal data-sharing policies ranked relatively low on the list of challenges (about 60% said it’s a moderate or significant challenge), discussions with CSPs indicate this may be a bigger problem than it seems. Many operators said that internal policies around data sharing are either inadequate or non-existent within their organizations, and the also said that this is a cultural problem.
Open APIs and a centralized framework for data distribution will go a long way toward solving the technical challenge and rendering the politics obsolete. In addition, CSPs’ senior leadership teams must push for creation of open data platforms that are accessible across their operating companies.
This is the approach Axiata took beginning in 2016 to address a significant gap that had resulted from having separate data analytics teams for each of its operating companies. These teams had deep technical knowledge but lacked business understanding, and business leaders were not leveraging data to help with business decisions. The company established Axiata Analytics in 2017, appointed Pedro Uria-Recio Vice President and Head of the division and used the TM Forum Big Data Analytics Solutions Suite to assess data maturity and implement global best practices for big data analytics in all the operating companies.
Read this case study to learn more about Axiata’s approach and watch for the new report, which will be published later this week.