| dc.description.abstract | Providing reliable communication to high-mobility users is one of the main challenges in next-generation wireless networks. The joint effects of high mobility and high data rate requirements result in doubly-selective wireless channels that are beyond the capabilities of orthogonal frequency division multiplexing (OFDM). To address these challenges, orthogonal time frequency space (OTFS) modulation as a foundational delay-Doppler multiplexing technique has emerged. In delay-Doppler multiplexing, a given transmit data symbol is spread across the entire time-frequency plane, which allows the exploitation of the full channel diversity gains. Thus, the time-varying multipath channel is transformed into a sparse and time-invariant channel in the delay-Doppler domain, where all received symbols experience approximately the same channel gain.
Although delay-Doppler domain multiplexing techniques have recently gained considerable attention for their ability to cope with time-varying channels, there is still a lack of investigation into their practical aspects. Pulse-shaping as a fundamental component of any practical communication system, prepares the signal for efficient transmission over the channel. Additionally, at the receiver, it is essential to identify the received signal accurately before proceeding with channel estimation and data detection. Consequently, synchronization is another critical aspect of practical communication systems, particularly in time-varying channels. Hence, this thesis investigates the practical aspects of the delay-Doppler multiplexing techniques, with a particular focus on pulse-shaping and synchronization.
The first part of this thesis introduces a unified framework that offers deep insights into the characteristics, similarities, and distinctions between circular and linear pulse-shaping methods. Utilizing this framework, a generalized input-output relationship is derived that captures the impact of pulse-shaping on the effective channel. This leads to the development of a unified modem for pulse-shaping on the delay-Doppler plane. The modem facilitates the design of fast, convolution-based low-complexity structures, which are significantly simpler than existing solutions in the literature. This structure reveals that while circular pulse-shaping techniques are less complex compared to linear methods, they are less effective in suppressing out-of-band (OOB) emissions. Consequently, two efficient techniques are proposed to mitigate the OOB emissions of circularly pulse-shaped signals. Specifically, these methods reduce OOB emissions by employing windowing and zero-guard (ZG) insertion along the delay dimension to smooth the signal edges.
After signal transmission, the received signal must be synchronized at the initial stage of the receiver. This process is typically achieved through a three-phase approach in current communication standards. The first phase occurs during downlink transmission, where each mobile terminal (MT) estimates timing offset (TO) and carrier frequency offset (CFO) using a pilot signal sent by the base station (BS). These estimated parameters are used by the MTs not only to detect the downlink data but also as reference points for synchronizing the uplink transmission. However, Doppler effects and propagation delays can cause the signals arriving at the BS to become outdated which results in residual synchronization errors in the uplink signals at the BS. Thus, the second phase involves estimating the TO and CFO in the uplink. In the final phase, the BS applies necessary corrections to these offsets and synchronizes the received signal.
As the first phase of the above process, this thesis investigates synchronization for downlink OTFS over a linear time-varying (LTV) channel. For this purpose, two estimators for obtaining TO and CFO in OTFS are proposed, using the same impulse pilot (IMP) signal employed for channel estimation. Specifically, the TO estimator identifies the start of each OTFS block by searching for a periodic sequence in both the delay and time dimensions. The CFO is then estimated by calculating the mean of the angles from the two-dimensional correlation function at the best timing instant. However, using an IMP is practically infeasible due to the large peak-to-average power ratio (PAPR). To tackle this issue, TO and CFO estimation techniques have been developed for a low-PAPR pilot structure with a cyclic prefix, known as PCP.
The TO estimator takes advantage of the periodic properties of the PCP in both the delay and time domains to identify the starting point of each OTFS block. Next, a two-stage CFO estimation technique is introduced. It first provides a coarse CFO estimate, which is then refined using a maximum likelihood (ML)-based approach proposed in this thesis. The ML-based method utilizes the generalized complex exponential basis expansion model (GCE-BEM), which absorbs the time-variation of the channel into the pilot and provides more accurate CFO estimation.
The final practical aspect addressed in this thesis focuses on the second and third stages of the synchronization process for the uplink in multiuser OTFS (MU-OTFS) systems under high-mobility scenarios.
This part investigates the accurate estimation and correction of the TOs and CFOs. Specifically, the estimated TOs are essential for locating the pilots of different users on the delay-time plane. Furthermore, the CFOs need to be estimated to achieve more accurate channel estimation. To achieve this, a spectrally efficient and practical pilot pattern is proposed, where each user transmits a PCP within a shared pilot region on the delay-Doppler plane. At the receiver, a bank of filters is utilized to separate different users' signals and accurately estimate their TOs and CFOs. Using a derived threshold range, a method is employed to yield precise TO estimates, enabling the identification of each user's pilot region. Subsequently, the CFO estimation technique reduces the complex multi-dimensional ML search to multiple one-dimensional search processes. This method leverages the Chebyshev polynomial-based basis expansion model (CPF-BEM) to effectively capture the time variation of the channel and provide accurate CFO estimates for all users. | en |
