Training, Testing and Validation Challenges for Next Generation AI/ML-Based Intelligent Wireless Networks

The ambition of providing ubiquitous, fast, and mobile communication services to all types of devices, such as handheld, vehicular and IoT, has resulted in tremendous complexity in our wireless networks, and the trend is only likely to accelerate. As a result, it is becoming exponentially harder to create good analytical models to describe the behavior of systems. Even if models exist, they are complex enough that optimization of such a system is prohibitive in terms of complexity. A few concrete examples of such optimization problems can help illuminate the issue better.