A year after NVIDIA released its Aerial wireless software and the Sionna simulation platform as open source projects, the ecosystem has seen rapid expansion. Global developer adoption of these platforms has grown tenfold, with more than 1 million downloads recorded for NVIDIA’s open source 5G and 6G tools. Over 8,000 research and academic institutions worldwide are now using these platforms to test ideas and compare results.
These tools connect simulation and AI training with experiments on real over-the-air wireless systems. Researchers are using them to build networks that do more than transmit data. They are creating systems that receive signals more effectively, keep autonomous systems connected in dynamic environments, and adapt in real time as devices move.
Semantic Communications for Robotics
One major challenge in next-generation networks is response time for robots and autonomous systems. When AI decisions run on a nearby server or in the cloud, the wireless connection becomes part of the total response time. Researchers are exploring goal-oriented semantic communications to solve this. Instead of sending full images, robots transmit compact descriptions of objects, their positions, and their relationships. This approach transmits only the information needed, allowing robots to respond faster in dynamic environments like warehouses.
At King’s College London, researchers connected a camera-equipped robot arm to an NVIDIA Jetson AGX Orin running NVIDIA Sionna. The system supports both software-defined 5G networks and edge AI processing. In experiments involving fruit packing and cube stacking with deliberately introduced faults, the approach reduced robot task completion time by up to 52.6 percent. It also improved task success rates compared to sending periodic images. The system proved resilient, detecting dropped objects and recovering from unsuccessful actions. In configurations using a digital twin, it checked planned movements for potential collisions before execution.
AI Receivers and Energy Efficiency
Interference and changing radio conditions often make wireless signals harder to decode, causing errors and retransmissions that delay information. AI receivers address this by learning to recover information from incoming signals more effectively. SoftBank demonstrated this by integrating a neural receiver into NVIDIA Aerial CUDA-Accelerated RAN (ACAR). In controlled tests with interference, the neural receiver delivered higher uplink throughput than a conventional receiver at all 15 tested settings, with a maximum gain of roughly 60 percent.
ETH Zurich researchers found that tuning neural receivers using local measurements reduced block error rates by more than one-third compared with selected pretrained versions. This improvement helps responders maintain a clearer view of emergency scenes, such as fires or search areas.
Managing energy demand is another critical area. Networks must handle peak loads, such as those in busy stadiums, but demand drops when crowds leave. AI can identify when spare capacity can power down to reduce electricity use without dropping connections. Researchers at the University of Leeds developed a system that predicts network demand 10 minutes ahead. It uses these forecasts to guide base station sleep decisions, checking that neighboring active stations can continue serving affected devices. Using a portable, NVIDIA DGX Spark-powered 5G network-in-a-box alongside a digital twin, the team reported a 45.6 percent reduction in RAN power consumption relative to a full-power baseline while maintaining device connectivity.
Accelerating Simulation with GPUs
Before deploying a network, researchers need to know how it will perform when hundreds of devices compete for connections. Simulating these conditions helps teams evaluate designs before committing to costly equipment. However, realistic simulations can take hours, limiting how many ideas researchers can explore. The Centre Tecnològic de Telecomunicacions de Catalunya (CTTC) is accelerating this process by bringing GPU-accelerated radio channel models, running on ACAR, into its open source 5G-LENA network simulator.
These models calculate how wireless signals travel between antennas and devices. In an early evaluation, the approach reduced simulation runtime from approximately three hours to 10 to 20 minutes, depending on the integration. This preserved the expected statistical behavior and network performance. Faster simulations let researchers compare more designs and investigate more demanding scenarios. This helps shorten the path from a proposed innovation to a thoroughly evaluated network design.
Researchers and developers will showcase more innovations shaping the future of wireless networks at NVIDIA GTC Washington, D.C., during 6G Developer Sessions.
Source: NVIDIA
