Article Dans Une Revue Studies in Engineering and Exact Sciences Année : 2024

Efficient LLR approximation using machine learning in symmetric-alpha-stable channels

Résumé

In environments characterized by impulsive noise, such as Symmetric-α-Stable (SαS) channels, the channel parameters α and γ need to be fixed to approximate the Log-Likelihood Ratio (LLR) computation required by many decoders. The possible values of γ are typically very close for a fixed value of α. In this paper, we demonstrate that for a fixed α, it is not necessary to calculate the LLR for every possible value of γ, since the variation in the LLR is negligible when the values of γ are close. Therefore, we use the LLR approximation for fixed α and γacross all values of γ in our simulation by selecting an intermediate value for γ. Consequently, it is only necessary to know the parameter α at the receiver, as γ does not significantly affect performance. This further simplifies the practical implementation of LLR approximation in environments characterized by impulsive noise. In our simulations, we compare the performance of the exact LLR of 4-ASK and its approximation under known channel parameters α and γ, using a single value of γ for a fixed α. Our findings confirm that the effect of γ on performance is negligible when its values are close, validating our choice of an intermediate value for γ. This approach simplifies the LLR approximation process while maintaining accuracy in performance evaluation.

Fichier principal
Vignette du fichier
689+Caderno+P.-1.pdf (396.9 Ko) Télécharger le fichier
Origine Fichiers éditeurs autorisés sur une archive ouverte
Licence
Commentaire Revue en Open Access sans précision de licence, mais avec autorisation explicite de rediffuser : https://ojs.studiespublicacoes.com.br/ojs/index.php/sees/Open_access

Dates et versions

hal-04879206 , version 1 (23-05-2025)

Licence

Identifiants

Citer

Latifa Mostari, Alban Goupil, Abdelmalik Taleb-Ahmed. Efficient LLR approximation using machine learning in symmetric-alpha-stable channels. Studies in Engineering and Exact Sciences, 2024, 5 (2), pp.e11822. ⟨10.54021/seesv5n2-689⟩. ⟨hal-04879206⟩
172 Consultations
146 Téléchargements

Altmetric

Partager

  • More