Communications and Signal Processing Lab (ComSP)

Title	Técnicas de adaptación del enlace mejoradas para redes celulares
State	Finished
Author	Francisco Blanquez Casado
Director/s	Mª Carmen Aguayo Torres ,  Gerardo Gómez Paredes
University	Universidad de Málaga
Center	Escuela Técnica Superior de Ingeniería de Telecomunicación
Department	Ingeniería de Comunicaciones
Reading date	01/06/2017
File	PDF

In this thesis, a study of the link adaptation process in LTE is carried out, proposing new techniques to improve its performance. This work is focused on two key parts: 1) the turbo decoding process and 2) the adaptive modulation and coding.

Turbo decoders based on Soft Output Viterbi Algorithm (SOVA) truncate the number of operations of the decoding process to a fix value to reduce complexity. In this work, we show how this truncation might degrade the decoding performance if the number of operation is too low, or might carry out unnecessary operations, and so increase power consumption, if it is too high. Furthermore, the appropriate number of operations depends on the transmission conditions, which can be time variant. An adaptive SOVA (adSOVA) is proposed in this thesis to dynamically adapt this number of operations. Then, the adSOVA is able whether to increase the number of operations to avoid degradation in error correction performance or to reduce it to save power without degrading performance, depending on the transmission conditions.

Adaptive modulation and coding process is usually supported by the well-known Outer Loop Link Adaptation (OLLA) technique, which is widely used to ensure a target BLER. However, there is a lack of analysis of the OLLA behavior in the literature. In this thesis a deep analysis of the OLLA is carried out. For this purpose, binary logistic regression has been used to model BLER when turbo coding is used. Results of this analysis are the convergence conditions of the OLLA, as well as the effect of the step sizes in its performance. From the previous analysis of the traditional OLLA, an enhanced OLLA (eOLLA) is proposed, which is able to adapt its step size according to the deviance of the target BLER. In addition to that, the eOLLA is able to track temporal variations of the channel when no transmissions are done, while the traditional OLLA needs transmissions to update its offset. Thus, the proposed eOLLA outperforms the traditional OLLA, especially in low load traffic scenarios. Additionally, binary logistic regression is used to obtain a full analytic model of the adaptive modulation and coding process with channel coding based on turbo coding for constant power in LTE downlink. This model shows that an optimal solution would require of an OLLA with multiple offsets, which would lead to improve the spectral efficiency as much as 20%. Finally, the influence of the variance of the transmitted data block size in adaptive modulation and coding is studied, showing how a high variance might degrade its performance. Then, a Closed Loop Link Adaptation (CLLA) technique is proposed, which is able to modify the modulation and coding scheme suggested by the UE depending on the size of the data block to be transmitted, in order to ensure the target BLER. The CLLA outperforms OLLA in scenarios with high variance of data block sizes