Runlength-limited (RLL) sequences have recently gained attention as a means to cope with the self-introduced intersymbol interference (ISI) in communications systems employing faster-than-Nyquist (FTN) signaling, in particular in systems using zero-crossing modulation (ZXM). However, previous research has revealed a significant gap between the achievable rates in ZXM systems using i) sequences generated by practical RLL codes and ii) maxentropic RLL sequences, which cannot solely be explained by the lower code rate of practical RLL codes. In this paper, we give evidence that the assignment between bit sequences and RLL sequences is essential for the performance of an RLL code when transmission over noisy channels is considered. We propose an RLL coding scheme with a high degree of freedom for the selection of the assignment and present a simple algorithm for its optimization. We show that our codes outperform RLL codes published in the literature in terms of the achievable rate in ZXM systems.

Energy efficiency is a critical challenge for next-generation mobile networks. Especially as traffic demand grows, the energy per bit must decrease significantly. One promising solution is the Gearbox-PHY, which adaptively switches between modulation schemes and tailored radio front ends to maximize energy efficiency while delivering required data rates. In this regard, high spectral efficiency needs can be addressed with standard quadrature amplitude modulation, while low-power alternatives like impulse radio are employed for lower data rate scenarios, significantly reducing front end power consumption. While we considered the energy optimization for such a Gearbox-PHY in prior work, the specific focus of this paper is the consideration of the trade-off between oscillator power consumption and its phase noise. Using literature-based models and measurements for hardware power consumption, we demonstrate that embracing hardware impairments can lead to substantial energy savings of up to three orders of magnitude.

Modulation schemes that encode information in the time domain play an important role for energy-efficient communications at high and low spectral efficiencies. Verdú showed that a spectrally efficient operation of pulse-based modulation schemes with independent identically distributed symbols at low signal-to-noise ratios is only possible when the phase is used as an additional degree of freedom. In the present work, four maxentropic symbol sequences are considered that encode information in the time domain and make use of the phase to varying degrees. The spectral efficiency is evaluated for the transmission over a bandlimited additive white Gaussian noise channel where the output is quantized with the lowest resolution that allows the reconstruction of the input symbols in the noiseless case. Moreover, faster-than-Nyquist (FTN) signaling is considered. Our results show that using the phase only to provide redundancy already leads to a significantly increased spectral efficiency, a lower required energy per communicated bit, and allows to benefit from FTN signaling. This makes the previously proposed zero-crossing modulation and time-derived zero-crossing modulation, a specific kind of impulse radio, to energy-efficient contenders for wideband millimeter-wave communications and for the operation of lowpower wireless sensor nodes in the Internet of things, respectively.

Utilizing 1-bit quantization at the analog-to-digital converter (ADC) is a promising approach to reduce the problematically high power consumption of high resolution ADCs in millimeter-wave (mmWave) and sub-terahertz (THz) communications. However, as 1-bit quantization is a highly nonlinear operation standard channel estimation algorithms cannot be applied. Therefore, we study algorithms for channel estimation in receivers with 1-bit quantization under consideration of a two-wave with diffuse power (TWDP) fading channel model, which was shown to be a realistic model for indoor communications in the mmWave regime. We combine maximum-likelihood (ML) amplitude estimation with a least-squares (LS) phase estimation approach known from literature to estimate the fading channel based on blocks of pilot symbols periodically inserted into the transmit symbol sequence. Furthermore, we apply Wiener filtering for interpolation of the channel estimates at the data blocks. The estimation performance of the proposed algorithms is evaluated numerically in terms of the mean squared error (MSE) and the suitability of the approach is demonstrated by evaluating the coded block error rate (BLER) for an exemplary system in comparison to the case with perfect channel knowledge. Our results show that almost the same BLER can be achieved by utilizing the derived estimation approach as compared to a system with perfect channel knowledge.

Digital receivers based on 1-bit quantization and temporal oversampling w. r. t. the transmit signal bandwidth are a promising solution for the design of energy-efficient communications systems in the millimeter-wave (mmWave) and sub-terahertz bands. However, off-the-shelf algorithms for channel estimation cannot be applied as 1-bit quantization is a highly non-linear operation. Phase noise (PN) in particular has a deteriorating effect on the communication performance at these high frequencies and, therefore, needs to be tracked and compensated at the receiver. In this context, we derive an analytical solution for a close approximation of the Bayesian Cramér-Rao bound for PN estimation in systems employing 1-bit quantization, which provides insights into the impact of various design parameters on the achievable estimation performance. Furthermore, we use the bound to benchmark the performance of two existing PN estimators, showing that one of these estimators performs close to the optimum.

Wireless communications systems beyond 5G are expected to utilize large available bandwidths at frequencies above 100 GHz in order to achieve data rates above 100 Gbit/s. However, the power consumption of the analog-to-digital converters (ADCs) for such systems is becoming a major challenge. Trading a reduced amplitude resolution for an increased temporal resolution by employing temporal oversampling w.r.t. the Nyquist rate is a possible solution to this problem. In this work, we consider a wireless communications system employing zero-crossing modulation (ZXM) and 1-bit quantization in combination with temporal oversampling at the receiver, where ZXM is implemented by combining runlength-limited (RLL) transmit sequences with faster-than-Nyquist (FTN) signaling. We compare the performance and complexity of four different soft-output equalization algorithms, namely, two approximations of the linear minimum mean squared error (LMMSE) equalizer, a BCJR equalizer and a deep-learning based equalizer, for such systems. We consider the mutual information (MI) between the input bits of the RLL encoder and the output log-likelihood ratios (LLRs) of the RLL decoder as a performance measure and evaluate it numerically. Our results demonstrate that one of the proposed LMMSE equalizers outperforms the competing algorithms in the low and mid signal-to-noise ratio (SNR) range, despite having the lowest implementational complexity.