Implementation and Evaluation of Semantic Communication on SDR Based LoRa Platform

In recent years, due to the rise of semantic communication in sixth generation (6G) wireless communication systems, the research on semantic transmission has gained increasing attention in Internet of Things (IoT). In this paper, we develop a software-defined radio (SDR) based long range(LoRa) communication platform, which utilize the universal software radio peripheral (USRP) and GNU Radio to implement an intelligent end-to-end semantic communication system, comprising two key levels: the Semantic Level and the LoRa Transmission Level. Specifically, we generate semantic vectors at the sending end by extracting the meaning of the data. In this case, the data transmitted by the LoRa physical layer is the semantic information with redundancy removed. We set the spreading factor (SF) as 7, the signal bandwidth as 250 kHz, the preamble length as 8 and the sampling rate as 1 MHz, under these configured LoRa parameters, we successfully transmit 10 sentences in a single transmission event. Experimental results demonstrate that at the receiving end, the bilingual evaluation understudy 1-grams (BLEU) score achieve 0.86, and the bidirectional encoder representations from transformers (BERT) similarity score achieve 0.96. This high-lights the exceptional performance of our intelligent end-to-end LoRa-Semantic communication system in conveying semantic information. Even when simultaneously transmitting multiple sentences, the receiver accurately comprehends the conveyed meanings.