CHANNEL ESTIMATION TECHNIQUES FOR FASTER-THAN-NYQUIST SIGNALING
Date
2025-05-12
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ORCID
0000-0001-6480-3425
Type
Thesis
Degree Level
Doctoral
Abstract
The rapid growth of wireless data traffic, driven by the increasing number of mobile devices and data-heavy applications, has created a significant demand for fast and reliable wireless communication systems. Since bandwidth remains a constrained resource, enhancing spectral efficiency (SE) is a key challenge. Faster-than-Nyquist (FTN) signaling has attracted significant interest as a technique to enhance the SE of wireless systems. This approach allows for data transmission rates faster than the Nyquist rate, without requiring additional bandwidth or increased energy per bit. However, FTN signaling introduces a significant challenge, particularly in channel estimation and data detection, because of inter-symbol interference (ISI) at the receiver, which becomes more pronounced in frequency-selective and doubly-selective (i.e., time- and frequency-selective) fading channels.
To facilitate accurate channel estimation in wireless systems, pilot symbols, which are predefined symbols known to both the transmitter and receiver, are commonly adopted. In the multiplexing pilot (MP) channel estimation method, time-division multiplexing involves inserting pilot symbols within the transmitted data stream. While this approach enables reliable channel estimation, it reduces the effective data rate and compromises the SE. To mitigate this drawback, the superimposed pilot (SP) approach has been proposed, where known pilot symbols are superimposed on the data stream. This approach overcomes the challenges of MP-aided methods, which suffer from substantial SE loss in high-mobility environments because of the requirement for more frequent pilot sequences embedded within the data. However, SP-aided methods come at the cost of introducing interference between the pilot and data symbols. Addressing the interference introduced by SP becomes especially challenging in the presence of ISI caused by FTN signaling. This thesis focuses on developing MP- and SP-aided channel estimation techniques for FTN signaling.
The first main contribution of this thesis is an MP-aided channel estimation method for FTN signaling over doubly-selective channels. This approach is built on the least sum of squared errors (LSSE) technique to estimate complex channel coefficients at pilot locations while effectively addressing the ISI caused by both FTN signaling and frequency-selective fading. The method includes the design of an optimal pilot sequence aimed at minimizing the mean square error (MSE) of channel estimation. To manage the time-selective nature of the channel, a low-complexity linear interpolation method is employed to track channel variations at data symbol locations. For FTN signaling data detection, a turbo equalizer utilizing a soft-input soft-output (SISO) MMSE algorithm is integrated. At the same spectral efficiency, simulation results demonstrate a notable enhancement—around 6 dB in channel estimation MSE and approximately 3.5 dB in the BER of FTN signaling—achieved by using the our method, in comparison to the approach of Ishihara and Sugiura (2017) over frequency-selective channels.
The second main contribution of this thesis is proposing a novel index modulation (IM)-based MP-aided channel estimation method tailored for FTN signaling in high-frequency (HF) communications. The method comprises two algorithms: pilot sequence placement (PSP) and pilot sequence location identification (PSLI). The PSP algorithm leverages the locations of pilot sequences to convey additional information, thereby enhancing SE. The PSLI algorithm enables the receiver to identify pilot sequence locations based on the specific characteristics of HF channels and the favorable autocorrelation properties of the optimal pilot sequence. Simulation results reveal that the proposed method achieves comparable MSE and BER performance in channel estimation to that of Keykhosravi and Bedeer (2023) over frequency-selective channels, while also providing improved SE. Moreover, the results demonstrate an enhancement in SE, along with approximately 6 dB improvement in the MSE of channel estimation and around 3.5 dB gain in the BER of FTN signaling, compared to the method proposed by Ishihara and Sugiura (2017) under the same channel conditions.
The third main contribution presents a novel SP-aided channel estimation method for FTN signaling over doubly-selective channels. The proposed frame structure superimposes periodic pilot sequences onto the data symbols, eliminating the overhead associated with MP methods and thereby enhancing SE. An optimal SP sequence is designed to minimize the MSE of channel estimation. To address the challenges posed by the time-varying behavior of the channel, we utilize a basis expansion model (BEM) to represent the time-varying channel tap weights as a combination of basis functions with time-invariant coefficients. Building on the proposed SP-aided estimation method, two distinct methods are developed: the SP-aided separate channel estimation and data detection (SCEDD) method, which employs a turbo equalizer for data detection, and the SP-aided joint channel estimation and data detection (JCEDD) method, which iteratively updates channel estimates during the equalization process to improve detection accuracy. Our simulation results demonstrate that, at the same SE, the proposed SP-aided SCEDD method for FTN signaling significantly outperforms the MP-aided approach presented in the works of Keykhosravi and Bedeer (2023), and Ishihara and Sugiura (2017) in terms of MSE and BER, particularly under higher fading rates on the order of 10^(-3). Furthermore, under identical SE conditions, the proposed SP-aided JCEDD method consistently achieves superior MSE and BER performance compared to these existing methods. Specifically, the techniques presented in Keykhosravi and Bedeer (2023), and in Ishihara and Sugiura (2017) are unable to effectively track rapid channel variations such as those observed in the practical high-latitude HF channel defined in the ITU-R F.1487 recommendation, characterized by a Doppler frequency of 10 Hz (fading rate of 0.004). In contrast, our SP-aided JCEDD method demonstrates reliable performance under these challenging conditions. Under very low fading rates—specifically for the practical low-latitude channel defined in the ITU-R F.1487 recommendations with a Doppler frequency of 1 Hz (fading rate of 0.0004)—the proposed SP-aided JCEDD algorithm achieves more than 2 dB and 6 dB improvements in MSE compared to the methods in Keykhosravi and Bedeer (2023), and Ishihara and Sugiura (2017), respectively. In terms of BER, while our method outperforms the MP-aided approach in Ishihara and Sugiura (2017) by over 3 dB, it maintains comparable performance to the MP-aided FTN system in Keykhosravi and Bedeer (2023), with less than 0.5 dB degradation.
Description
Keywords
Channel estimation, faster-than-Nyquist (FTN) signaling, least sum of squared errors (LSSE), doubly-selective channels, minimum-mean square error (MMSE), soft-input soft-output (SISO) equalization, index modulation (IM), Basis expansion model (BEM), superimposed pilot (SP)
Citation
Degree
Doctor of Philosophy (Ph.D.)
Department
Electrical and Computer Engineering
Program
Electrical Engineering