Reduced-complexity Digital Predistortion in Flexible Radio Spectrum Access
Abdelaziz, Mahmoud (2017)
Abdelaziz, Mahmoud
Tampere University of Technology
2017
Teknis-taloudellinen tiedekunta - Faculty of Business and Technology Management
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:ISBN:978-952-15-3998-5
https://urn.fi/URN:ISBN:978-952-15-3998-5
Tiivistelmä
Wireless communications is nowadays seen as one of the main foundations of technological advancements in, e.g., healthcare, education, agriculture, transportation, computing, personal communications, media, and entertainment. This requires major technological developments and advances at different levels of the wireless communication systems and networks. In particular, it is required to utilize the currently available frequency spectrum in a more and more efficient way, while also adopting new spectral bands. Moreover, it is required that cheaper and smaller electronic components are used to build future wireless communication systems to facilitate increasingly cost-effective solutions. Meanwhile, energy efficiency becomes extremely important in wide scale deployments of the networks both from a running cost point of view, and from an environmental impact point of view. This is the big picture, or the so called ‘bird’s eye view’ of the challenges that are yet to be met in this very interesting and fast developing field of science.
The power amplifier (PA) is the most power-hungry component in most RF transmitters. Consequently, its energy efficiency significantly contributes to the overall energy efficiency of the transmitter, and in fact the whole wireless network. Unfortunately, energy efficiency enhancement implies operating the PA closer to its saturation region, which typically results in severe nonlinear distortion that can deteriorate the signal quality and cause interference to neighboring users, both of which negatively impact the system spectral efficiency. Moreover, in flexible spectrum access scenarios, which are essential for improving the spectral efficiency, particular in the form of non-contiguous radio spectrum access, the nonlinear distortion due to the PA becomes even more severe and can significantly impact the overall network performance. For example, in noncontiguous carrier aggregation (CA) in LTE-Advanced, it has been demonstrated that in addition to the classical in-band distortion and regrowth around the main carriers, harmful spurious emission components are generated which can easily violate the spurious emission limits even in the case of user equipment (UE) transmitters.
Technological advances in the digital electronics domain have enabled us to approach this problem from a digital signal processing point of view in the form of widely-adopted and researched digital predistortion (DPD) technology. However, when the signal bandwidth gets larger, and flexible or non-contiguous spectrum access is introduced, the complexity of the DPD increases and the power consumed in the digital domain by the DPD itself becomes higher and higher, to the extent that it might be close to, or even surpass, the energy savings achieved from using a more efficient PA. The problem becomes even more challenging at the UE side which has relatively limited computational capabilities and lower transmit power. This dilemma can be resolved by developing novel reduced-complexity DPD solutions in such flexible spectrum access and/or wide bandwidth scenarios while not sacrificing the DPD performance, which is the main topic area that this thesis work contributes to.
The first contribution of this thesis is the development of a spur-injection based sub-band DPD structure for spurious emission mitigation in noncontiguous transmission scenarios. A novel and effective learning algorithm is also introduced, for the proposed sub-band DPD, based on the decorrelation principle. Mathematical models of the unwanted emissions are formulated based on realistic PA models with memory, followed by developing an efficient DPD structure for mitigating these emissions with reducedcomplexity in both the DPD main processing and learning paths while providing excellent spurious emission suppression. In the special case when the spurious emissions overlap with the own RX band in frequency division duplexing (FDD) transceivers, a novel subband DPD solution is also developed that uses the main RX for DPD learning without requiring any additional observation RX, thus further reducing the DPD complexity.
The second contribution is the development of a novel reduced-complexity concurrent DPD, with a single-feedback receiver path, for carrier aggregation-like scenarios. The proposed solution is based on a simple and flexible DPD structure with decorrelationbased parameter learning. Practical simulations and RF measurements demonstrate that the proposed concurrent DPD provides excellent linearization performance, in terms of in-band error vector magnitude (EVM) and adjacent channel leakage ratio (ACLR), when compared to state-of-the-art concurrent DPD solutions, despite its reduced computational complexity in both the DPD main path processing and parameter learning.
The third contribution is the development of a new and novel frequency-optimized DPD solution which can tailor its linearization capabilities to any particular regions of the spectrum. Detailed mathematical expressions of the power spectrum at the PA output as a function of the DPD coefficients are formulated. A Newton-Raphson optimization routine is then utilized to optimize the suppression of unwanted emissions at arbitrary pre-specified frequencies at the PA output. From a complexity reduction perspective, this means that for a given linearization performance at a particular frequency range, an optimized and reduced-complexity DPD can be used.
Detailed quantitative complexity analysis, of all the proposed DPD solutions, is performed in this thesis. The complexity and linearization performance are also compared to state-of-the-art DPD solutions in the literature to validate and demonstrate the complexity reduction aspect without sacrificing the linearization performance. Moreover, all the DPD solutions developed in this thesis are tested in practical RF environments using real cellular power amplifiers that are commercially used in the latest wireless communication systems, both at the base station side and at the mobile terminal side. These experiments, along with the strong theoretical foundation of the developed DPD solutions prove that they can be commercially used as such to enhance the performance, energy efficiency, and cost effectiveness of next generation wireless transmitters.
The power amplifier (PA) is the most power-hungry component in most RF transmitters. Consequently, its energy efficiency significantly contributes to the overall energy efficiency of the transmitter, and in fact the whole wireless network. Unfortunately, energy efficiency enhancement implies operating the PA closer to its saturation region, which typically results in severe nonlinear distortion that can deteriorate the signal quality and cause interference to neighboring users, both of which negatively impact the system spectral efficiency. Moreover, in flexible spectrum access scenarios, which are essential for improving the spectral efficiency, particular in the form of non-contiguous radio spectrum access, the nonlinear distortion due to the PA becomes even more severe and can significantly impact the overall network performance. For example, in noncontiguous carrier aggregation (CA) in LTE-Advanced, it has been demonstrated that in addition to the classical in-band distortion and regrowth around the main carriers, harmful spurious emission components are generated which can easily violate the spurious emission limits even in the case of user equipment (UE) transmitters.
Technological advances in the digital electronics domain have enabled us to approach this problem from a digital signal processing point of view in the form of widely-adopted and researched digital predistortion (DPD) technology. However, when the signal bandwidth gets larger, and flexible or non-contiguous spectrum access is introduced, the complexity of the DPD increases and the power consumed in the digital domain by the DPD itself becomes higher and higher, to the extent that it might be close to, or even surpass, the energy savings achieved from using a more efficient PA. The problem becomes even more challenging at the UE side which has relatively limited computational capabilities and lower transmit power. This dilemma can be resolved by developing novel reduced-complexity DPD solutions in such flexible spectrum access and/or wide bandwidth scenarios while not sacrificing the DPD performance, which is the main topic area that this thesis work contributes to.
The first contribution of this thesis is the development of a spur-injection based sub-band DPD structure for spurious emission mitigation in noncontiguous transmission scenarios. A novel and effective learning algorithm is also introduced, for the proposed sub-band DPD, based on the decorrelation principle. Mathematical models of the unwanted emissions are formulated based on realistic PA models with memory, followed by developing an efficient DPD structure for mitigating these emissions with reducedcomplexity in both the DPD main processing and learning paths while providing excellent spurious emission suppression. In the special case when the spurious emissions overlap with the own RX band in frequency division duplexing (FDD) transceivers, a novel subband DPD solution is also developed that uses the main RX for DPD learning without requiring any additional observation RX, thus further reducing the DPD complexity.
The second contribution is the development of a novel reduced-complexity concurrent DPD, with a single-feedback receiver path, for carrier aggregation-like scenarios. The proposed solution is based on a simple and flexible DPD structure with decorrelationbased parameter learning. Practical simulations and RF measurements demonstrate that the proposed concurrent DPD provides excellent linearization performance, in terms of in-band error vector magnitude (EVM) and adjacent channel leakage ratio (ACLR), when compared to state-of-the-art concurrent DPD solutions, despite its reduced computational complexity in both the DPD main path processing and parameter learning.
The third contribution is the development of a new and novel frequency-optimized DPD solution which can tailor its linearization capabilities to any particular regions of the spectrum. Detailed mathematical expressions of the power spectrum at the PA output as a function of the DPD coefficients are formulated. A Newton-Raphson optimization routine is then utilized to optimize the suppression of unwanted emissions at arbitrary pre-specified frequencies at the PA output. From a complexity reduction perspective, this means that for a given linearization performance at a particular frequency range, an optimized and reduced-complexity DPD can be used.
Detailed quantitative complexity analysis, of all the proposed DPD solutions, is performed in this thesis. The complexity and linearization performance are also compared to state-of-the-art DPD solutions in the literature to validate and demonstrate the complexity reduction aspect without sacrificing the linearization performance. Moreover, all the DPD solutions developed in this thesis are tested in practical RF environments using real cellular power amplifiers that are commercially used in the latest wireless communication systems, both at the base station side and at the mobile terminal side. These experiments, along with the strong theoretical foundation of the developed DPD solutions prove that they can be commercially used as such to enhance the performance, energy efficiency, and cost effectiveness of next generation wireless transmitters.
Kokoelmat
- Väitöskirjat [4866]