Transmissão de imagens em canais de HF.

Abstract

The long range without repeaters is on the reasons of strategic importance of communcation based on High Frequency band (HF). In special to military communications, this importance increases as the possibility of to stablish a link in a war scenario. However, due to its low bandwith and severe channel conditions, HF channels are normally used to voice and low data rate transmission. In spite of the well stablished standards, to increase transmission possibilities, like image transmission, new techniques have to be implemented. Image transmission in the HF channel is a challenge, and to perform this task it is necessary to pre-process the image, e.g. by compressing it with a high rates followed by a powerful error correction encoding. An alternative is to use a joint source/channel encoding, and one fo these techniques uses vector quantization combined adjusted to the channel modulation. In this work a previously proposed vector quantization combined with modulation is extended by reduction the modulation order from 16-QAM to 4-QAM, more adequated to the HF channel. The proposed modification was performed by training a dictionary in a Khohonen self organized map in a topology with 8 dimensions. Computational simulation results show the robustness of the proposed modification even in low signal-to-noise regime.