Neural network verification improves deep joint source channel coding robustness
Neural Network Verification for Deep Joint Source-Channel Coding
Software EngineeringArtificial Intelligence
Summary
Sending images over wireless channels can be tricky because small changes like noise or interference can drastically reduce the image quality when using neural networks to compress and transmit data. The authors created a new way to check and guarantee how bad the worst-case errors can be when images are sent this way, even when signal conditions change. They improved existing verification tools to handle special parts of their neural network and modeled wireless effects more realistically, making their checks more precise. Their approach also works well on real wireless hardware, confirming the safety bounds they found in practice.
What this means in practice
- •For wireless communication engineers: Certify and improve the reliability of neural decoding in image transmission under realistic channel disturbances using verified robustness bounds.
- •For software-defined radio developers: Validate and deploy deep joint source-channel coding models with guaranteed performance in real wireless hardware environments.
Authors
Thanh Le, Hai Duong, Takeshi Matsumura, ThanhVu Nguyen
Abstract
Deep joint source-channel coding (DeepJSCC) transmits data end-to-end over wireless channels using a neural encoder-decoder, but reconstruction quality can degrade sharply under adversarial perturbations and channel disturbances; no method formally bounds this degradation for DeepJSCC. We present the first bound-propagation framework for verifying DeepJSCC's decoder, bounding worst-case reconstruction error over a given wireless channel's noise region. Current deep neural network (DNN) verifiers do not support three DeepJSCC decoder components: parametric rectified linear activations (PReLU), transposed convolutions, and Rayleigh fading. We extend state-of-the-art techniques for optimization of linear relaxation in DNN verification for PReLU, replace the transposed convolution with its restricted upsample-then-convolution form, and formulate Rayleigh fading as a structural perturbation prepended directly into the decoder, thereby reducing the dimensionality of the verification problem. We also instantiate Lipschitz-regularized global robustness training, denoted GloRo, improving global robustness and enabling tight certification of DeepJSCC models for the first time. On DeepJSCC model for image transmission, this global robustness training procedure combined with structural encoding lowers the median certified bound by up to 41% and certifies about ten times more safe cases (192 against 19) than GloRo with interval encoding at a 10-degree error in channel estimation. Over-the-air validation with an orthogonal frequency-division multiplexing (OFDM) implementation on software-defined radio devices confirm the certificate holds on real hardware, with a worst observed error on radio link at 0.082 against a certified bound of 0.128.