A Study on Signal Decomposition and Polarization-Division Multiplexing in Communication System for Power Amplifier Efficiency Enhancement
Year: 2024
Type: dissertation
Abstract: This dissertation introduces polarization-division multiplexing (PDM) communication systems aimed at enhancing power amplifier (PA) efficiency. The proposed PDM systems undergo signal decomposition of original baseband signals, resulting in decomposed signals with Peak-to-Average Power Ratios (PAPRs) generally lower than the original signal. These signals are then separately amplified through dedicated PAs and subjected to PDM. Since each PA amplifies signals with reduced PAPR through decomposition, the overall PA efficiency in this system is enhanced. However, the transformation of the RF system introduces cross-polarization interference (XPI) resulted from PDM. PDM involves multiplexing different data into distinct polarization states. In PDM, complete separation of the multiplexed data is hindered by interference between polarization states, known as XPI. This interference leads to deteriorated error vector magnitude (EVM) at the Rx. Therefore, signal processing techniques are applied for demodulation at the Rx to mitigate EVM deterioration caused by XPI. \nThe primary focus of this dissertation is on two modulation schemes: single carrier modulation (SCM) and orthogonal frequency division multiplexing (OFDM). Signal decomposition methods at the Tx for different modulation schemes are proposed, and corresponding Rx demodulation methods are suggested. The dissertation provides a comprehensive analysis of the proposed system's performance, presenting a reduction in total DC power consumption within the PAs while aligning the Symbol Error Rate (SER) with that of the original system. Performance analyses consider two channels: Additive White Gaussian Noise (AWGN) channel and frequency-selective Rayleigh fading channel. Additionally, a prototype operating at a 28 GHz carrier frequency is fabricated for experiments to verify the efficiency enhancement feasibility. Comparative analysis based on experimental results assesses the overall performance between the proposed and original systems. \nIn SCM, a high-order Quadrature Amplitude Modulation (QAM) is decomposed into two asymmetric low-order QAMs, which are subjected to PDM. A technique for recovering the EVM which is deteriorated by XPI, enabled by the asymmetric decomposition, is proposed and applied. In the assumption of frequency-selective Rayleigh fading channel, the proposed system yields an average DC power reduction ratio of 54%/67%/72% (compared to original 16/64/256 QAM, respectively) when the optimized balanced scaling is applied for the two asymmetric low-order QAMs while it becomes 40%/52%/51% (compared to original 16/64/256 QAM, respectively) when balanced scaling is not applied. Furthermore, the experimental results show a DC power reduction ratio of 32.5%/43%/43.4% (compared to original 16/64/256 QAM, respectively) with no balanced scaling, while exhibiting improved SER. \nIn OFDM, the original signal is decomposed into sub-signals with reduced effective PAPR of 5.4 and 5.9 dB based on a predefined threshold, which will be subjected to PDM. An EVM recovery technique enabled by sparsity of one of the sub-signals is proposed and filters are applied after channel equalization to mitigate noise amplification in channels with poor quality. The proposed system induces an average DC power consumption reduction of 46% in PAs compared to the original 16 QAM in the assumption of frequency-selective Rayleigh fading channel. Additionally, experimental results using a prototype demonstrate the feasibility of efficiency enhancement showing EVM performance comparable to that of the original system. \nIn OFDM with suppressed out-of-band emission (OOBE), the original signal undergoes decomposition based on a predefined threshold, and the resulting sub-signals have significant OOBE. OOBE of them are suppressed and after that, they undergo PDM. An EVM recovery technique based on sparsity of one of the sub-signals (before the OOBE-suppression was applied) is proposed and applied. It involves solving a compressive sensing problem after a filter is applied in channel equalization to reject noise amplification at the channel with poor quality. The proposed system induces an average DC power consumption reduction of 36% in PAs compared to the original 16 QAM in the assumption of frequency-selective Rayleigh fading channel. Additionally, experimental results using a prototype demonstrate the feasibility of efficiency enhancement showing EVM performance comparable to that of the original system. \nFinally, the scalability of the proposed system to multiple-input and multiple-output (MIMO) is briefly addressed. In the context of MIMO scenarios, it is investigated whether replacing a portion of the RF chains connected to each Tx antenna with the proposed system and employing the suggested signal decomposition for PA DC power reduction would still yield advantages by MIMO in Tx performance.
Author: 김형철
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7034181363
