Networks are costly from planning to deployment. They need to be more efficient to see a return on investment for 6G, which is expected to deliver these higher efficiencies. Until now, operators have experienced several significant advances, including the shift to cloud-native 5G standalone and the disaggregation of hardware and software in the RAN to help contain costs while expanding their monetization possibilities. Everything changed with AI.
Viavi Solutions Inc., for example, created its 6G Forward Program, specializing in wireless technology research and testing. Creating 6G proofs of concept using its GPU-powered real-time 6G testbed, advances include new waveforms in the MIMO delay-Doppler domain, advanced multiple-access technology that addresses overloaded networks and full AI neural network transmission.
Because 6G represents the first AI-native generation of wireless communications, research is currently focused on large-scale RF propagation channel modeling based on AI and ML technologies. It is a collaborative effort between Viavi and several industry players, as well as the Institute for the Wireless Internet of Things and the Open6G cooperative research center at Northeastern University, to develop a business case for 6G.
A 6G city-scale digital twin
Viavi’s 6G and AI research within the program is now yielding results. For example, its collaborative efforts have resulted in the creation of a 6G city-scale digital twin that plays an important role in training components such as the 6G AI-Native Air Interface, which is fundamentally changing how communication technologies are developed and deployed.
A digital twin uses AI, ML, sensor data and real-time 3D to create a digital simulation that mimics real-world network conditions and facilitates in-depth testing across a variety of scenarios. Using AI algorithms and closed-loop automation, AI-based digital twins solve complex, real-world challenges faster and with less risk.
What does that look like? For one thing, it replaces climbing cell towers. Instead, drones capture images, entering them into a database and then into a large-language Markov model tuned and optimized for image classification. A digital twin of the operator’s cell tower allows an operator with a VR headset to correlate whether a tower’s antennas are offline due to storms or other factors. Generative AI and digital twins can also monitor and fix data centers by creating 3D renderings, making it possible to measure elements such as heat and power consumption remotely.
The addition of AI and its integration into the network and the cloud, as well as for services and testing, created challenges and opportunities. AI-enabled network optimization, monitoring and troubleshooting are improving efficiency and availability. The challenge now is that all of it must be tested, and both AI models and test methodologies must also be standardized. To that end, Viavi’s neural receivers are used in a comprehensive test and training setup integrating AI-assisted design and advanced modulation techniques.
The AI-Native Air Interface
The goal of the AI-Native Air Interface is to effectively supply data to applications while simultaneously addressing communication problems and hardware limitations. The neural receiver concept will replace traditional signal-processing blocks with trained AI models.
For comprehensive testing of neural receivers, Viavi’s receiver is part of an end-to-end test and training environment. The base station uses an AI-aided constellation design, replacing the modulation block with a neural network to design a custom constellation for the receiver. The receiver is generalizable over a variety of channel models and modulations; delay and doppler; signal-to-noise ratio; and conditions that are important for deployment in practical systems. The company uses open-source and in-house libraries to implement the transceiver architecture.
Viavi’s 6G research is important in the creation of networks that dynamically adapt to different environments and conditions. Northeastern University’s Institute for the Wireless Internet of Things uses its testing and measurement equipment to model and deploy high-fidelity digital twins of real-world wireless networks on an Open RAN Digital Twinning platform, Colosseum. This facilitates the development and experimentation of AI-driven solutions for networks in a risk-free digital-twinned world.
The Viavi platform is now enabling operators to run simulations of large events, including concerts or autonomous vehicles driving on streets. The trained models are then deployable.
The company is also using AI and ML to augment ray tracing for propagation modeling in a digital twin that provides greater accuracy for operators to optimize networks. This modeling is key for wireless operators to simulate path loss, coverage and interference of a network configuration.