Short Polar Codes for Bluetooth
Master thesis
Permanent lenke
https://hdl.handle.net/11250/3096559Utgivelsesdato
2023Metadata
Vis full innførselSamlinger
Sammendrag
Denne studien undersøker anvendelsen av Polar-koder i sammenheng med Bluetooth, ved å se på bruken av Polar-koder med korte blokklengder, og sammenlikne den med konvolutionell kode under samme betingelser. Hvor bra de forskjellige dekoderene, CRC-aided List Viterbi Algorithem og CRC-aided Successive Cancellation List (SCL), presterer i å korriger overføringsfeil blir studert.
Simuleringer ble kjørt i MATLAB under varierende forhold, og resultatene indikerte at Polar-koder har bedre feilkorrigering enn konvolusjonskoder for bestemte blokklengder, og at forbedringer kan oppnås ved å øke liste-størrelsen. Studiet understreker potensialet for Polar-koder i å forbedre ytelsen innen Bluetooth, samtidig som behovet for videre forskning på flere faktorer som kompleksitet og varierende kodningsrater påpekes. Polar codes have gained significant attention in academia and industry as the first provably capacity-achieving forward error correcting code family with provable encoding and decoding complexity. Their significance is underscored by their inclusion in the new generation wireless communication standards, 5G. While initially designed for large block lengths, the 5G standards have shed light on the possibility to use Polar codes also with short block lengths. This study aims to investigate the suitability of Polar codes for enhancing data transmission efficiency in Bluetooth by comparing their performance with convolutional codes, as used in Bluetooth, under the same conditions.
This research addresses several sub-goals, including understanding the error-correction functionality of polar codes and convolutional codes, evaluating the performance of the Successive Cancellation List (SCL) decoder and the List Viterbi Algorithm (LVA) decoder in terms of Frame-Error-Rate (FER), assessing the role of the Cyclic Redundancy Check (CRC) as a supportive element in list decoding, and conducting a comparative analysis between the CRC-aided SCL and CRC-aided LVA decoders.
Simulations were run in MATLAB with various conditions, including different list sizes L and frame lengths K, using both a 24-bit and 6-bit CRC. The Binary Phase-Shift Keying (BPSK) modulation scheme over an Additive White Gaussian Noise (AWGN) channel was chosen and different signal-to-noise ratio (SNR) values were used to plot the FER. The findings of this research indicates that polar code have better error-correction then convolution code for block lengths including and larger than K=64 bits where 24 of them are dedicated to CRC-bits. Furthermore, increasing the list size improves the performance of both polar codes and convolutional codes. The use of 6-bit CRC instead of 24-bit CRC introduced false positives, making the overall code performance slightly worse, but increasing the throughput by reducing the redundant bits.
The results highlight the advantages of Polar codes over convolutional codes, particularly for larger frame sizes, and provide insights into the impact of CRC usage. While this study shows Polar codes have potential in improving data transmission efficiency in Bluetooth, it is important to consider additional factors such as complexity, varying code rates, and techniques like puncturing and shortening in future research. Overall, this investigation offers valuable insights for optimizing communication systems and showcases the potential of Polar codes in enhancing performance within short-range communication applications like Bluetooth.