Task-Oriented Communication for Graph Data: A Graph Information Bottleneck Approach

Graph data, essential in fields like knowledge representation and social networks, often involves large networks with many nodes and edges. Transmitting these graphs can be highly inefficient due to their size and redundancy for specific tasks. This paper introduces a method to extract a smaller, task-focused subgraph that maintains key information while reducing communication overhead. Our approach utilizes graph neural networks (GNNs) and the graph information bottleneck (GIB) principle to create a compact, informative, and robust graph representation suitable for transmission. The challenge lies in the irregular structure of graph data, making GIB optimization complex. We address this by deriving a tractable variational upper bound for the objective function. Additionally, we propose the VQ-GIB mechanism, integrating vector quantization (VQ) to convert subgraph representations into a discrete codebook sequence, compatible with existing digital communication systems. Our experiments show that this GIB-based method significantly lowers communication costs while preserving essential task-related information. The approach demonstrates robust performance across various communication channels, suitable for both continuous and discrete systems.

Semantic Communications with Ordered Importance using ChatGPT

This letter proposes a novel semantic communication scheme with ordered importance (SCOI) using the chat generative pre-trained transformer (ChatGPT). In the proposed SCOI scheme, ChatGPT plays the role of a consulting assistant. Given a message to be transmitted, the transmitter first queries ChatGPT to output the importance order of each word. According to the importance order, the transmitter then performs an unequal error protection transmission strategy to make the transmission of essential words more reliable. Unlike the existing semantic communication schemes, SCOI is compatible with existing source-channel separation designs and can be directly embedded into current communication systems. Our experimental results show that both the transmission bit error rate (BER) of important words and the semantic loss measured by ChatGPT are much lower than the existing communication schemes.