Voice FMS is a signaling-based platform for retail & wholesale operators and enterprises intending to protect their outbound/inbound traffic against fraudulent activities such as International Revenue Share Fraud, PBX Hacking, Wangiri Calls and Callbacks, Wangiri 2.0, False Answer Supervision, Robocalls, Spam/Scam Calls, etc.
The platform is available as a cloud-based solution and on client’s premises. All protocols of signaling data transmission (SIP, HTTP, RADIUS, DIAMETER, etc.) are supported.
Adjustable blocking, alerting, and reporting policies allow you to choose how to respond to different fraud events. Along with email notifications, you can activate additional channels such as Telegram, Slack, Mattermost and HTTP for receiving alerts and reports.
If you would like to activate additional notification channels, please contact your manager.
FMS operates a signaling-based system, analyzing data from active calls in real time. As soon as its high-speed computational algorithms automatically detect patterns of fraudulent traffic, the client is immediately alerted, and a predetermined action is instantly performed.
Both Behavioral Analysis and Proactive Monitoring methods utilize unique approaches of fraud analysis based on machine learning (deep neural networks) and mathematical modeling, developed entirely in-house. This enables FMS to detect not only well-disguised fraud attacks but also novel traffic anomalies and fraud types (using unsupervised machine learning). A notable example of a discovered fraud scenario is Wangiri 2.0.
FMS is used by different types of clients (enterprises, mobile and landline operators, transit carriers). Due to this, it is trained to detect a wide array of fraud types, identify new fraud scenarios and tendencies to fraud sophistication.
Due to the advantages mentioned above, FMS demonstrates exceptional accuracy: an almost complete absence of undetected or late-detected attacks and an extremely low false positive rate:
Accuracy > 99.8%
Accuracy = (TP + TN) / (TP + TN + FP + FN)
It is a proportion of correct predictions among the total number of events
Precision > 99%
Precision = TP / (FP + TP)
Total number of events, correctly flagged as fraudulent, relative to total number of events, flagged as fraudulent
Recall > 97%
Recall = TP / (FN + TP)
Total number of events, correctly flagged as fraudulent, relative to total number of fraudulent events
False Positive Rate (FPR) < 0.01%
FPR = FP / (FP + TN)
Total number of events, falsely flagged as fraudulent, relative to total number of non-fraudulent events
TP - true positives; TN - true negatives; FP - false positives; FN - false negatives
Retail traffic is extremely precious to each and every client: service denial is entirely unacceptable in this context. The system should avoid setting unnecessary or excessive blocks of number ranges. This is why a special approach to enterprise traffic protection is provided in FMS:
Security Operations Centre (SOC) is a group of fraud analysts and support engineers which work 24/7. SOC support can be provided to clients as a dedicated service. This service includes:
If you would like to activate SOC service, please contact your manager.
Flashcalls are zero-duration calls, which contain PIN-codes as the last digits of A-numbers. They are not quite fraud calls, but can be undesirable for carriers/terminating operators, because flashcalls (as zero-duration calls) do not bring them any revenue. With FMS cutting-edge technology clients can identify flashcalls for further monetization or block them as unwanted call attempts.
If you would like to activate flashcalls blocking feature, please contact your manager.
FMS does not require a CapEx solution or significant time and resources spent on integration. Installation and setup of the system can be done remotely, enabling fast out-of-the-box deployment. Quick implementation of the system allows for immediate commencement of traffic protection.
The AI engine detects the preparation of fraudulent activity in voice traffic. By establishing well-designed blocking and alerting policies for each fraud type, the client can prevent fraud attacks before they happen, avoiding even minor fraud-related damages. Note that calls rejected by FMS receive a SIP response code set to 603. To modify this, please contact your manager.
Our advanced machine learning algorithms detect even well-disguised fraud attacks on hubbing & retail traffic in real-time. After the parameters of fraudulent traffic are determined, the pre-configured blocking actions for each fraud type allow the client to eliminate fraud with minimal possible losses.
The client can configure any real-time threshold-based rules to monitor usage and end-customers’ behavior. Once the threshold is reached, the incident is alerted to the client. All corresponding active calls can be broken, and related traffic can be blocked for a predefined period.
The client may blocklist or allowlist both A&B number ranges. Block List is the first priority rule. If the range is blocklisted, it means that there will be no further investigation for this range. All the calls from/to blocklisted ranges will be rejected without any analysis (the information about such calls can be found in Dashboard). Allow List is the second priority rule. If the range is allowlisted, it means that there will be no further investigation for this range. All the calls from/to allowlisted ranges will go through without any analysis.