PAIQoO: Quantifying the Network Requirements of Physical AI
PAIQoO quantifies the network requirements of Physical AI applications by characterizing how sensor fidelity, end-to-end latency, and radio conditions jointly determine task performance. Higher-fidelity sensing improves performance only until the additional data begins to delay the action, and a slightly sub-optimal action executed on time often outperforms a better action that arrives late. The goal is to translate task-level requirements into concrete network requirements, so that a network-aware sensing model can select what to sense, and at what fidelity, as conditions change.