Voice Control System
The security management and control system of voice dedicated line adopts two-level management and control architecture, realizing the separation of "management" and "control". Centralized management platform focuses on centralized management, analysis and traceability, while call control platform focuses on traffic control and information collection.

The centralized management platform realizes unified access management, unified policy configuration, unified harassment analysis, and unified audit and inspection for the whole network, forming the centralized management and control ability of voice private line, centralized management and management of voice line business, centralized configuration of policy rules, analysis of harassment phone behavior, centralized analysis, audit, distribution and control strategy of illegal content, and visual display and analysis of operation reports.
The call management and control platform adopts real-time traffic control scheme for management and control to "side span". In this way, ISBC, AGCF and MGCF are connected to the core network. Through the real-time processing of full signaling and some media data (flexible sampling), and uploading the bill, signaling and recording files to the centralized management platform, the call traffic is controlled according to the unified strategy, and the whitelist is authenticated.
The analysis and identification ability of harassing calls is to receive xDR bills, signaling and logs from the call management and control platform, analyze and read data records, and build an analysis model of harassment calls; identify harassment calls through detection models such as machine learning, and dynamically adjust and optimize the existing voice control strategy.

Using big data analysis technology and machine learning technology, through data collection and cleaning, feature construction, auxiliary unsupervised clustering analysis, automatic iteration and other functional modules, the harassment phone analysis model is formed, which can be self iterated and has strong applicability.




