|
Summary iSpectra is a real-time platform for collecting, analyzing, and managing mobile network interference, fully compliant with international telecommunications standards. The system integrates AI/ML to identify various characteristic interference sources and performs intelligent geolocation using heatmaps, transforming hundreds of thousands of BTS sites into a distributed sensing network without requiring additional hardware investment. In its first quarter of operation, iSpectra detected 600 interference-affected sites - an increase of 100% compared to 2024. |
Developed by
VIETTEL SOFTWARE · A subsidiary of Viettel Group - Vietnam’s Military Industry and Telecommunications Group
The rapid expansion of 4G and 5G networks in Vietnam - with more than 120,000 BTS sites across Viettel, VNPT, and MobiFone, and over 600,000 cells covering 98% of the population. This has created an extremely dense radio frequency environment.
In this context, radio interference is unavoidable. This directly impacts service quality and the experience of millions of users.
Introduction to iSpectra
1.1. Bottlenecks of the traditional interference handling process
The traditional interference handling process presents three critical bottlenecks:
• Completely reactive: The process is only initiated when operators detect service degradation - at which point interference may have persisted for days or even weeks.
• Manual and data-deficient: The entire workflow: from statistics, analysis to coordination between operators and regulators - is conducted via Excel files and direct communication, with no visualization and no analytical history.
• Lack of cross-network visibility: Each operator only sees its own data, making it impossible to distinguish between external interference and internal faults, leading to misdirected investigations and significant resource waste.
1.2. Traditional interference handling workflow
The traditional interference handling process consists of six steps, all based on manual coordination between operators and the regulator:
Traditional interference handling process
|
Step |
Responsible Party |
Description |
|
1 |
Operator |
Compile affected BTS sites via Excel reports - slow updates, large volume, prone to errors. |
|
2 |
Operator |
Dispatch field engineers to inspect equipment. If no internal fault is found, proceed to the next step. |
|
3 |
Operator → Regulator |
Submit reports to the regulator (Authority of Radio Frequency Management - ARFM) for support. |
|
4 |
Regulator |
Analyze Excel data to define measurement areas, coordinate via email/phone - fully manual. |
|
5 |
Regulator (Field) |
Conduct field measurements using specialized equipment to locate interference sources. |
|
6 |
Repeat |
The entire process repeats until interference is resolved - often taking weeks or months. |
|
Current Situation Data is aggregated via Excel - slow updates and difficult analysis. Predictive analysis relies entirely on individual experience. Field engineers must perform multiple measurements due to lack of directional information. The entire process can take weeks to months for complex interference cases, while service quality continues to degrade. |
1.3. Urgent needs
Three stakeholder groups require a unified solution:
|
Regulator |
Operators |
End Users |
|
Enhance proactive management Require nationwide real-time monitoring of the interference floor. Enable data-driven decision-making instead of relying on reports. Support policymaking for mobile network development. |
Optimize coverage and service quality Detect previously unknown internal interference. Reduce time and cost for field measurements. Improve service quality for customers. |
Protect citizen
health Reduce radio frequency (RF) energy absorption by the human body, as user equipment (UE) does not need to increase transmission power to compensate for interference. |
|
Required Solution A shared system that enables regulators to proactively monitor and operators to proactively resolve interference - based on centralized data instead of fragmented experience.. |
||
Critical needs of the telecommunications industry
2. Solution: iSpectra
iSpectra is a real-time system for collecting, analyzing, and managing mobile network interference, developed by Viettel Software during 2023–2025.
The core philosophy of iSpectra is fundamentally different from existing solutions. Instead of building new measurement infrastructure, iSpectra repurposes over 300,000 active BTS sites as a massive distributed RF sensing network. It continuously collects data automatically, without human intervention and without additional hardware investment.
iSpectra acts as a “central coordination and processing hub” for interference data, enabling operators and regulators to shift from reactive to proactive management.
2.1. Core Features
Below are these main features of iSpectra:
|
Feature 1 |
Value Delivered |
|
Automated data collection Directly integrates with each operator’s OSS via standardized APIs. Enables continuous, automated data collection and data exchange. |
Data Assurance: - Accurate: Monitoring interference KPIs directly instead of indirect QoS KPIs - Complete: Full data coverage around interference locations down to PRB level - Clean: Direct extraction of raw data - Real-time: Hourly granularity, daily updates |
|
Feature 2 |
Value Delivered |
|
Automated analysis Analyze geolocation and interference floor data using AI models, grouping nearby sites affected by the same interference source into clusters for simultaneous processing. |
Eliminate internally affected sites (caused by operator equipment faults). Enable simultaneous multi-site processing to improve operational efficiency. |
|
Intelligent localization Analyze deviations in interference KPIs combined with antenna configuration data to visualize the predicted location of interference sources on Google Maps. |
Reduce field measurement and interference source search time. |
|
Feature 3 |
Value Delivered |
|
Automated classification and orchestration Based on analysis results, automatically assign tasks to responsible units. Automatically set deadlines and monitor KPIs in real time. |
Ensure processing timelines and service quality commitments to operators. |
|
Closed-loop workflow Automatically receive and return results to stakeholders via APIs, email, and notification portal. |
The entire process (from detection, analysis, task assignment to completion confirmation) is managed on a single platform. No need for external communication via email or phone. |
2.2. AI/ML Technology: Intelligent Analysis and Localization
This is the most technically distinctive aspect of iSpectra. By applying AI and Machine Learning to automatically identify interference source types and locate their positions, iSpectra replaces experience-based judgment with data-driven analysis.
a. Interference source identification using convolutional neural networks
The system utilizes GIS, graph analytics, and Convolutional Neural Networks (CNN) to recognize interference spectrum patterns. The process consists of 7 steps:
• Collect interference spectrum data from affected BTS sites (real-world data).
• Analyze spatial topology to identify neighboring sites potentially affected by the same source
• Evaluate similarity to identify and cluster cells with similar characteristics
• Apply Convolutional Neural Networks to classify interference source types based on signal characteristics and network parameters
• Based on CNN predictions, generate the final result as clusters of sites affected by the same interference source
• Use site geolocation data, differences in interference signal strength between cells, and antenna configuration to estimate the source location
• After field verification, feed validated data back into the training dataset to improve the model
iSpectra has identified and classified various characteristic types of interference sources, including:
|
No. |
Type |
Cause |
|
1 |
Access Point |
Unauthorized or faulty Wi-Fi devices emitting into mobile bands |
|
2 |
Guard Band (ICI) |
Misconfigured guard bands causing inter-channel interference |
|
3 |
Remote Radio Unit Error |
Hardware failure in Remote Radio Unit |
|
4 |
Cell Overshooting |
Overlapping coverage between nearby BTS sites |
|
5 |
External Repeater |
Illegal signal boosters |
|
6 |
Wireless Micro |
Unauthorized wireless microphone usage |
b. Intelligent Geolocation using AI and heatmaps
After classification, iSpectra applies AI-based triangulation using multi-site interference intensity data to generate heatmaps, identifying probable interference source locations. This allows technicians to target precise areas instead of random field searching.
3. Implement Results
3.1. Achievements
iSpectra serves as a powerful supporting tool that enables telecom operators to resolve incidents more quickly. As a result, it enhances user experience and ensures network stability in the context of rapidly growing mobile data demand.
|
600 interference-affected sites detected +100% vs 2024 |
160 interference sources identified +50% vs 2024 |
76% analysis accuracy up to 65% |
-30% reduction in call drop rate VNPT pilot |
3.2. Impact by Stakeholder
|
Regulator |
Operators (Viettel, VNPT, MobiFone) |
|
Proactive nationwide monitoring Detect interference even before operators become aware. The number of detected interference sources increased by 50%, processing is 30% faster and 15% more accurate. Enable data-driven decision-making instead of paper-based reporting. |
Breakthrough operational efficiency Interference resolution time reduced by 50%, equivalent to 1,080 hours per year. Detect previously overlooked internal interference (PIM, RRU faults). Reduce field operation and labor costs. |
Mobile usersService quality is significantly improved. Reduce call drop rates and data interruptions. Lower RF energy absorption as user devices do not need to increase transmission power to compensate for interference. |
National digital infrastructureEnsure connectivity quality for over 100 million users. Establish a national spectrum data foundation to support 5G and 6G policy planning. No additional investment in specialized measurement equipment is required. |
4. Comparison with other solutions
iSpectra is benchmarked against the reference baseline: the traditional manual approach (currently widely used globally). The detailed comparison is presented in the table below:
|
Tính năng |
iSpectra |
Phương pháp thủ công |
|
Automated BTS data collection |
✓ Yes |
✗ No |
|
Multi-operator integration |
✓ Yes |
✗ No |
|
Centralized analytics |
✓ Yes |
✗ No |
|
Heatmap & geospatial analysis |
✓ Yes |
✗ No |
|
ML-based classification |
✓ Yes |
✗ No |
|
AI-based localization |
✓ Yes |
✗ No |
|
3D spectrum analysis |
✓ Yes |
✗ No |
|
Workflow integration |
✓ Yes |
✗ No |
|
Internal/external classification |
✓ Yes |
✗ No |
|
Data security |
✓ Yes |
✓ Yes |
|
Competitive Advantage iSpectra’s core strength lies in multi-operator data sharing agreements, requiring legal, technical, and organizational coordination between regulators and competing operators. This creates an extremely high barrier to entry that no country - including developed nations - has yet overcome. At AWG-35, countries such as Japan, South Korea, China, and India highly appreciated the model but have not been able to implement a similar solution. |
5. iSpectra’s international recognition
- At AWG-35 (Asia-Pacific Telecommunity Wireless Group, Thailand, 9/2025), with 421 delegates from 21 countries, 47 enterprises, and 2 international organizations, iSpectra was recognized as a breakthrough solution.
- At AWG-35, Vietnam was officially assigned to lead research on AI/ML-based radio interference detection and control solutions for the Asia-Pacific region, based on the iSpectra model.
- Many APT member countries and leading global telecommunications equipment vendors have expressed strong interest in learning from and adopting this implementation model.
- Dr. Le Van Tuan, Director General of the Authority of Radio Frequency Management (ARFM) and the architect of iSpectra, was awarded the title of Honorary Chair by the AWG conference, in recognition of his six years of leadership in regional radio cooperation.
|
Long-term mission From a national system, iSpectra aims to become a reference model for AI-driven spectrum management across the Asia-Pacific region, where many countries face rapid growth of wireless devices without effective management tools. |
|
Why iSpectra deserves international recognition iSpectra solves a global challenge: integrating real-time data from multiple competing operators into a unified analytical system, transforming existing infrastructure into a national monitoring tool, and applying AI to classify interference sources with high accuracy. This is not an experiment - it is a live national-scale system protecting connectivity for over 100 million users, with measurable results from the first quarter of operation. This is the first successful deployment of its kind globally, recognized at AWG-35, and a testament to Vietnam’s capability to build world-class digital infrastructure. |
Viettel Software · Building National Digital Infrastructure · Made in Vietnam