In this work, we compare emergent communication (EC) built upon multi-agent deep reinforcement learning (MADRL) and language-oriented semantic communication (LSC) empowered by a pre-trained large language model (LLM) using human language. In a multi-agent remote navigation task, with multimodal input data comprising location and channel maps, it is shown that EC incurs high training cost and struggles when using multimodal data, whereas LSC yields high inference computing cost due to the LLM's large size. To address their respective bottlenecks, we propose a novel framework of language-guided EC (LEC) by guiding the EC training using LSC via knowledge distillation (KD). Simulations corroborate that LEC achieves faster travel time while avoiding areas with poor channel conditions, as well as speeding up the MADRL training convergence by up to 61.8% compared to EC.
@article{arxiv.2401.12624,
title = {Knowledge Distillation from Language-Oriented to Emergent Communication for Multi-Agent Remote Control},
author = {Yongjun Kim and Sejin Seo and Jihong Park and Mehdi Bennis and Seong-Lyun Kim and Junil Choi},
journal= {arXiv preprint arXiv:2401.12624},
year = {2024}
}