Adaptive Generative Models for Digital Wireless Channels

Salih, Omar; Wang, Cheng-Xiang; Ai, Bo; Mesleh, Raed

doi:10.1109/twc.2014.2325028

Cited by 7 publications

(5 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[6] and [7]). The most current approaches aim to design models that could be parameterized adaptively in a bit-by-bit fashion and be able to capture faster rate of change [8].…”

Section: Related Workmentioning

confidence: 99%

Modeling Poisson Error Process on Wireless Channels

Csoka

Polec

2022

Int. j. commun. netw. inf. secur.

View full text Add to dashboard Cite

Burst error modeling has seen extensive research and progress over several decades evolving into ever more complex modeling techniques used today. This paper analyzed usefulness of the most prominent generative and descriptive (analytical) methods. Data containing error bits and packets from real wireless transmission has been used to obtain statistical information about error burst and gap behavior in the channel and various generative and descriptive modeling techniques were applied to model the error process with the goal of establishing advantages and disadvantages of each technique. Generative methods were represented by the commonly implemented Elliot’s model with parameters calculated using a generalized algebraic form and descriptive methods were represented by one of the most flexible exponentially shaped distributions with regard to parameterization and heavy-tailed function modeling - gamma distribution, and lastly a technique represented by Markov modulated Poisson process (MMPP-2) producing second-order hyper-exponentially distributed characteristics. Results of the experiments were highly in favor of Elliot’s and MMPP-2 model demonstrating possible application of MMPP-2 model in application to commonly observed exponentially-shaped error process on the wireless channel.

show abstract

“…[6] and [7]). The most current approaches aim to design models that could be parameterized adaptively in a bit-by-bit fashion and be able to capture faster rate of change [8].…”

Section: Related Workmentioning

confidence: 99%

Modeling Poisson Error Process on Wireless Channels

Csoka

Polec

2022

Int. j. commun. netw. inf. secur.

View full text Add to dashboard Cite

show abstract

“…In particular, the generative models are probabilistic, driven by unsupervised learning approaches. Currently, there are three major types of generative models [33], [34], which are variational auto encoders (VAEs) [35], [36], generative adversarial networks (GANs) [37], [38] and normalizing flows (NFlows) [39]- [41]. Recently, the deep generative models are adopted in literature to solve the linear inverse problems [42].…”

Section: A Related Researchmentioning

confidence: 99%

Learning based signal detection for MIMO systems with unknown noise statistics

Fan

et al. 2021

Preprint

View full text Add to dashboard Cite

This paper aims to devise a generalized maximum likelihood (ML) estimator to robustly detect signals with unknown noise statistics in multiple-input multiple-output (MIMO) systems. In practice, there is little or even no statistical knowledge on the system noise, which in many cases is non-Gaussian, impulsive and not analyzable. Existing detection methods have mainly focused on specific noise models, which are not robust enough with unknown noise statistics. To tackle this issue, we propose a novel ML detection framework to effectively recover the desired signal. Our framework is a fully probabilistic one that can efficiently approximate the unknown noise distribution through a normalizing flow. Importantly, this framework is driven by an unsupervised learning approach, where only the noise samples are required. To reduce the computational complexity, we further present a low-complexity version of the framework, by utilizing an initial estimation to reduce the search space. Simulation results show that our framework outperforms other existing algorithms in terms of bit error rate (BER) in non-analytical noise environments, while it can reach the ML performance bound in analytical noise environments. The code of this paper is available at https://github.com/skypitcher/manfe.

show abstract

“…The generative error models have proposed based on three widely used generative models, namely, the Simplified Fritchman Model (SFM), the Baum-Welch based Hidden Markov Model (BWHMM), and the Deterministic Process Based Generative Model (DPBGM), and by considering factors such as the detection threshold, parameterizations and parallel mapper [16]- [18]. They proved the effectiveness of their proposed models by showing the well-matched characteristics when comparing the reference error statistics of the underlying descriptive models with those of the generative models with respect to both the burst error statistics and BER performance of the coded digital wireless transmission system.…”

Section: Related Workmentioning

confidence: 99%

“…For the binary channel models, the channel is characterized using error statistics in terms of burst errors or error clusters. The performance of binary channels is evaluated using the bit error rate (BER) or packet error rate (PER) [16]- [18]. For the performance evaluation of the binary data transmission system, the study of the underlying burst error process and the exploration of the statistical dependencies among errors are important prerequisites.…”

Section: Introductionmentioning

confidence: 99%

Modeling and Analysis of Error Process in 5G Wireless Communication Using Two-State Markov Chain

Myint

Sato

2019

IEEE Access

View full text Add to dashboard Cite

In fifth-generation wireless communications, data transmission is challenging due to the occurrence of burst errors and packet losses that are caused by multipath fading in multipath transmissions. To acquire more efficient and reliable data transmissions and to mitigate the transmission medium degradation in the 5G networks, it is important to study the error patterns or burst the error sequences that can provide insights into the behavior of 5G wireless data transmissions. In this paper, a two-state Markov-based 5G error model is investigated and developed to model the statistical characteristics of the underlying error process in the 5G network. The underlying 5G error process was obtained from our 5G wireless simulation, which was implemented based on three different kinds of modulation methods, including QPSK, 16QAM, and 64QAM, and was employed using the LDPC and TURBO coding methods. By comparing the burst or gap error statistics of the reference error sequences from the 5G wireless simulations and those of the generated error sequences from the two-state Markov error model, we show that the error behaviors of the coded OFDM 5G simulations can be adequately modeled by using the two-state Markov error model. Our proposed two-state Markov-based wireless error model can help to provide a more thorough understanding of the error process in 5G wireless communications and to evaluate the error control strategies with less computational complexity and shorter simulation times. INDEX TERMS 5G, burst error statistics, two-state Markov model, wireless error model.

show abstract

Adaptive Generative Models for Digital Wireless Channels

Cited by 7 publications

References 38 publications

Modeling Poisson Error Process on Wireless Channels

Modeling Poisson Error Process on Wireless Channels

Learning based signal detection for MIMO systems with unknown noise statistics

Modeling and Analysis of Error Process in 5G Wireless Communication Using Two-State Markov Chain

Contact Info

Product

Resources

About