Token-based control improves image data transmission over wireless channels
Thinking in Tokens, Talking in Bits: A Practical Interface for Token Communication
Information Theory
Summary
Modern artificial intelligence systems work by breaking data into small pieces called tokens, but communication networks usually send raw bits. The authors found a way for tokens to guide how actual bits are sent over the network, rather than sending tokens themselves, which can cause compatibility problems. They tested this idea using a system that classifies images and improves how the image data is compressed and protected during transmission. Their approach keeps the actual data sent as plain bits but uses tokens to decide how to manage those bits, resulting in better speed and accuracy when sending images over wireless channels.
What this means in practice
- •For wireless communication engineers: Improve image classification over noisy wireless channels by dynamically adjusting bit compression and error protection controlled by AI token relevance.
- •For digital video transmission teams: Optimize video and image streaming by controlling compression rates and error correction using AI-derived priorities without changing the bitstream format.
Authors
Chanho Park, Bumsu Park, Soonhee Kwon, Sangrim Lee, Namyoon Lee
Abstract
Advanced artificial intelligence models think in tokens; contemporary communication systems carry bits. The direct way to bridge this gap is to transmit tokens, but that makes a model-specific representation part of the air interface, coupling the endpoints through a shared tokenizer, codebook, and often a neural transceiver. We take a different route: keep bits in the payload and let tokens control how those bits are generated and protected. The resulting token-bit interface transition aligns task-side tokens with source- and channel-coding units, translates token relevance into codec controls, and preserves the induced priority order across the coding chain. We instantiate it for image classification, where a vision transformer scores the task relevance of each image region from its attention maps: those scores steer block-wise JPEG rate allocation, then group the compressed bits for protection at different polar-code rates. The payload remains an explicit, reconstructable bitstream recovered by a correspondingly configured decoder. Over-the-air experiments on a software-defined radio testbed show improved accuracy--latency tradeoffs over separate source-channel coding, performance competitive with far more memory-intensive neural joint source-channel coding, and graceful degradation under channel mismatch. Token communication, then, need not transmit tokens explicitly; what it needs is an interface through which tokens determine how bits are communicated.