HintKD: Hint Knowledge Distillation for Bandwidth-Constrained Cloud-Edge Inference
Abstract
Collaborative inference between the edge cloud and user equipment (UE) is a promising paradigm for deploying large deep neural networks (DNNs) in 6G networks. However, existing cloud-edge inference and distillation schemes often require the real-time transmission of high-dimensional intermediate features or soft probability vectors, which imposes a substantial communication burden on bandwidth-limited wireless links. To address this challenge, we propose HintKD, a bandwidth-constrained distillation framework for cloud-edge inference. The core idea is to compress the teacher's guidance into compact discrete hints rather than transmit raw features directly. Specifically, a cloud-side teacher is first trained with a maximum conditional mutual information (MCMI) objective to preserve informative intra-class variations. Its latent representation is then mapped to a compact codebook through differentiable vector quantization. During deployment, the cloud transmits only a hint index, and the UE uses a lightweight FiLM-based adapter to modulate student features according to the received codeword. Experiments on the DeepSense 6G beam prediction task show that HintKD preserves competitive accuracy while reducing the inference-side communication payload from O(D_t) for feature transmission or O(C) for logit transmission to O(1) for discrete hint transmission. In addition, experiments on the MNIST dataset further verify the generality of the proposed distillation method across different datasets and tasks. These results show that HintKD achieves a favorable accuracy-bandwidth trade-off by replacing high-dimensional feature or logit transmission with a compact discrete hint index.
Related
- Federated Language Models Under Bandwidth Budgets: Distillation Rates and Conformal Coverage
- Personalized Federated Learning through Global Knowledge Distillation and Local Head Adaptation
- A Statistical Perspective on Knowledge Distillation: Foundations, Classical Methods, and Large Language Model Extensions
- Behavioral Knowledge Distillation for Dynamic Graph Anomaly Detection in Trustworthy Power Grid Dispatching
- EnerGNN: Learning Optimization-Compatible Energy Functions for Exact Constrained Combinatorial Inference
BibTeX
@misc{luo2026hintkd,
title = {{HintKD: Hint Knowledge Distillation for Bandwidth-Constrained Cloud-Edge Inference}},
author = {Wanling Luo and Jiayi Zhang and Bokai Xu and Ziheng Liu},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.08412},
eprint = {2610.08412},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.08412},
}
RIS
TY - UNPB
TI - HintKD: Hint Knowledge Distillation for Bandwidth-Constrained Cloud-Edge Inference
AU - Wanling Luo
AU - Jiayi Zhang
AU - Bokai Xu
AU - Ziheng Liu
PY - 2026
DA - 2026/10/06
PB - arXiv
DO - 10.48550/arxiv.2610.08412
UR - https://arxiv.org/abs/2610.08412
AB - Collaborative inference between the edge cloud and user equipment (UE) is a promising paradigm for deploying large deep neural networks (DNNs) in 6G networks. However, existing cloud-edge inference and distillation schemes often require the real-time transmission of high-dimensional intermediate features or soft probability vectors, which imposes a substantial communication burden on bandwidth-limited wireless links. To address this challenge, we propose HintKD, a bandwidth-constrained distillation framework for cloud-edge inference. The core idea is to compress the teacher's guidance into compact discrete hints rather than transmit raw features directly. Specifically, a cloud-side teacher is first trained with a maximum conditional mutual information (MCMI) objective to preserve informative intra-class variations. Its latent representation is then mapped to a compact codebook through differentiable vector quantization. During deployment, the cloud transmits only a hint index, and the UE uses a lightweight FiLM-based adapter to modulate student features according to the received codeword. Experiments on the DeepSense 6G beam prediction task show that HintKD preserves competitive accuracy while reducing the inference-side communication payload from O(D_t) for feature transmission or O(C) for logit transmission to O(1) for discrete hint transmission. In addition, experiments on the MNIST dataset further verify the generality of the proposed distillation method across different datasets and tasks. These results show that HintKD achieves a favorable accuracy-bandwidth trade-off by replacing high-dimensional feature or logit transmission with a compact discrete hint index.
ER -