Semantic Forwarding for Next Generation Relay Networks
Enes Arda, Emrecan Kutay, Aylin Yener
TL;DR
This work tackles efficient text transmission in a three-node cooperative relay by introducing semantic-forwarding techniques that do not rely on channel state information. It proposes two transformer-based schemes, SLF and SPF, that build and utilize a dynamic semantic state at the relay via attention to either decode current tokens or predict next tokens, respectively. Evaluations over AWGN and Rayleigh channels using the Europarl corpus show that both schemes outperform a semantic-agnostic baseline in BLEU and semantic similarity, with SPF offering near-parallel performance to SLF while enabling proactive forwarding. The results indicate that learned semantic forwarding can provide robust, low-latency, content-aware communication for next-generation networks, and open avenues for multi-relay and task-oriented semantic cooperation.
Abstract
We consider cooperative semantic text communications facilitated by a relay node. We propose two types of semantic forwarding: semantic lossy forwarding (SLF) and semantic predict-and-forward (SPF). Both are machine learning aided approaches, and, in particular, utilize attention mechanisms at the relay to establish a dynamic semantic state, updated upon receiving a new source signal. In the SLF model, the semantic state is used to decode the received source signal; whereas in the SPF model, it is used to predict the next source signal, enabling proactive forwarding. Our proposed forwarding schemes do not need any channel state information and exhibit consistent performance regardless of the relay's position. Our results demonstrate that the proposed semantic forwarding techniques outperform conventional semantic-agnostic baselines.