INTRODUCTION

Telecommunication companies face huge challenges. They need to plan efficient transmitter and receiver placement while managing complex environments. Cloud RF, a global leader in radio frequency planning, faced similar issues.

Their simulations of the signal propagation and tower placement were slow and limited in scope. Running detailed calculations on terrain models consumed time and computing resources.

At TechnoLynx, we understood these struggles. We accepted the challenge to create a solution that improves speed, accuracy, and deployment flexibility. Our approach combined advanced programming, deep system understanding, and scalable design.

THE PROBLEM

Cloud RF had ambitious goals, while their existing system did not meet growing demands. Signal propagation simulations became too slow for large datasets. Real-time analysis was nearly impossible when handling detailed maps and modelling complicated urban layouts.

Moreover, tower placement algorithms require high computational power. Identifying optimal positions for transmitters and receivers on complex terrain slowed operations considerably.

Additionally, running simulations across different client environments led to inconsistencies. The platform-dependent design restricted Cloud RF’s ability to offer services universally. It limited the market reach and created frustration for users in varied environments.

THE SOLUTION

We designed a multi-layer solution with performance, flexibility, and scalability at its core. Every step focused on optimising signal propagation modelling and tower placement simulation.

Advanced Signal Propagation Engine

The first step was improving the handling of signal propagation. We designed a high-performance, multi-threaded engine. This component used height maps and voxel-based calculations to manage terrain data more efficiently.

A custom work-package scheduler controlled the task distribution across multiple processing units. This ensured that signal propagation calculations happened as quickly as possible.

CUDA-optimised algorithms added another layer of speed. We implemented both Cartesian and polar coordinate methods, which addressed different use cases while keeping computations efficient.

Validation was also critical; hence, slower reference algorithms were included. These performed ray casting in basic form, allowing developers to confirm accuracy when needed.

Tower Placement Optimisation

Optimising tower placement was a key task. We created advanced algorithms designed to identify optimal tower locations more intelligently. Rather than simply selecting a single tower location, our solution generated heatmaps displaying candidate areas where towers are best suited to cover target regions. These heat maps revealed zones offering the greatest coverage if a tower were placed there.

This approach shifted focus from single-point recommendations to visualising best placement areas. Once the heatmaps were generated, we used Monte Carlo methods to test thousands of configurations and determine the best patterns for coverage.

The biggest performance boost was up to 300x faster, generated by our proprietary algorithms creating the heatmaps efficiently, rather than just multithreading. These optimisations were essential, especially when processing large maps or intricate terrain.

Indoor Signal Propagation

Recognising the need for indoor accuracy, we also developed a voxel-based ray casting system. This component is powered by CUDA acceleration, working in tandem with point cloud to voxel conversion, improving the fidelity of indoor signal simulations.

Signal propagation in indoor environments posed another challenge. Walls, objects, and other obstacles can disrupt transmission, making simulations complex. We developed a CUDA-accelerated ray casting solution in voxel space. This allowed for accurate modelling of signals as they bounced, reflected, or were absorbed within buildings.

Point cloud conversion was integrated into the system, enabling the conversion of scanned building data into voxel maps, making indoor signal propagation modelling more detailed and realistic.

Deployment and Platform Support

Our solution needed to work everywhere; thus, we performed the following to ensure smooth deployment:

• We containerised the entire system using Docker, ensuring seamless deployment in any environment. No matter where the simulation ran, results stayed accurate and stable.

• We powered our cross-platform build system with CMake, which enabled compilation for all common operating systems. This removed dependency headaches and simplified installation for Cloud RF’s clients.

• We added thorough unit and integration tests to ensure stability and correctness across the platforms by verifying every change.

THE RESULTS

GPU Optimisation and Proprietary Logic Combined to Transform CloudRF’s System

The performance improved drastically, helping them offer faster and more precise simulations to customers:

• Propagation simulations ran up to 200-300% faster thanks to the optimised work-package scheduling and CUDA enhancements.

• Large height map processing accelerated up to 10-20x with GPU support.

• Tower placement heatmap generation became roughly 300x faster, with the largest improvement driven by proprietary algorithm design rather than only threading.

• Deployment became simple and consistent across client systems.

• Platform independence allowed the solution to reach new market segments.

These enhancements turned CloudRF’s new system into a core asset, providing superior performance and accuracy in the telecommunications infrastructure market.

Long-Term Value and Industry Impact

Beyond speed improvements, the project deepened CloudRF’s understanding of signal propagation and tower placement logic. Our reports and workshops helped their engineering teams better understand how propagation behaves across different terrains and scenarios.

The GPU-optimised, cross-platform system also prepared CloudRF for future expansion. As new hardware and technologies emerge, the flexible architecture makes future updates and enhancements straightforward.

This new foundation positions CloudRF to continue leading in advanced RF planning solutions.

Enhancing Future Capabilities with AI Integration

Looking ahead, Cloud RF stands to benefit even further by integrating AI into the platform. While signal propagation and tower placement algorithms already provide excellent performance, AI offers opportunities to refine predictions and adapt to changing conditions.

Machine learning models can help identify patterns in data that traditional methods may miss. For example, analysing real-world signal feedback could improve predictions for specific terrains or weather conditions. This would make the system even more reliable in varied environments.

AI could also assist in optimising tower placement by learning from past deployment successes and failures. A smart algorithm could propose adjustments based on regional usage patterns or known interference issues. This reduces the manual effort required from network planners.

Another area for AI is automating adjustments in real time. As conditions change, such as new buildings being constructed, the AI-driven system could adapt signal propagation calculations accordingly. This ensures users always receive the best possible recommendations without needing manual updates.

By combining AI with the strong foundation already built, Cloud RF can make its signal propagation and tower placement system smarter and more adaptive. TechnoLynx remains ready to support these future steps with advanced machine learning expertise and tailored solutions.

WHY TECHNOLYNX?

At TechnoLynx, we understand how critical signal propagation and tower placement are. Our team specialises in optimising complex simulations using the latest AI and GPU-based technologies.

Whether your business needs tower placement simulation, signal propagation modelling, or advanced computing solutions, we can help. Our engineers will work closely with you to design systems which are fast, accurate, and easy to use.

From signal propagation algorithms to real-time analysis tools, our solutions are always built for performance and future scalability.

Contact TechnoLynx today to see how we can bring speed, accuracy, and efficiency to your RF planning and optimisation tasks.

Read more about our successful projects: Case-Study: Text-to-Speech Inference Optimisation on Edge