Communication System Design and Analysis for Asynchronous Molecular Timing Channels

Farsad, Nariman; Murin, Yonathan; Guo, Weisi; Chae, Chan-Byoung; Eckford, Andrew W.; Goldsmith, Andrea

doi:10.1109/tmbmc.2018.2885288

Cited by 19 publications

(19 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Under CSI uncertainty, the Viterbi algorithm implements Algorithm 1 with the loglikelihoods computed using the noisy channel estimate, while ViterbiNet is trained using samples taken from different realizations of the noisy h(γ). [41], underwater acoustic channels [42], and impulsive noise channels [43]. In particular, we simulate an alphastable noise with stability parameter α = 0.5, skewness parameter β = 0.75, scale parameter c = 1, and location parameter µ = 0, following […”

Section: A Time-invariant Channelsmentioning

confidence: 99%

ViterbiNet: A Deep Learning Based Viterbi Algorithm for Symbol Detection

Shlezinger

Farsad

Eldar

et al. 2020

IEEE Trans. Wireless Commun.

Self Cite

126

130

View full text Add to dashboard Cite

Symbol detection plays an important role in the implementation of digital receivers. In this work, we propose ViterbiNet, which is a data-driven symbol detector that does not require channel state information (CSI). ViterbiNet is obtained by integrating deep neural networks (DNNs) into the Viterbi algorithm. We identify the specific parts of the Viterbi algorithm that are channel-model-based, and design a DNN to implement only those computations, leaving the rest of the algorithm structure intact.We then propose a meta-learning based approach to train ViterbiNet online based on recent decisions, allowing the receiver to track dynamic channel conditions without requiring new training samples for every coherence block. Our numerical evaluations demonstrate that the performance of ViterbiNet, which is ignorant of the CSI, approaches that of the CSI-based Viterbi algorithm, and is capable of tracking time-varying channels without needing instantaneous CSI or additional training data. Moreover, unlike conventional Viterbi detection, ViterbiNet is robust to CSI uncertainty, and it can be reliably implemented in complex channel models with constrained computational burden. More broadly, our results demonstrate the conceptual benefit of designing communication systems to that integrate DNNs into established algorithms.Parts of this work were accepted for presentation in the

show abstract

Section: A Time-invariant Channelsmentioning

confidence: 99%

ViterbiNet: A Deep Learning Based Viterbi Algorithm for Symbol Detection

Shlezinger

Farsad

Eldar

et al. 2020

IEEE Trans. Wireless Commun.

Self Cite

126

130

View full text Add to dashboard Cite

show abstract

“…This random noise has been unveiled as a Lévy distribution [18], [54]- [56], an inverse Gaussian distribution [1], [6], a stable distribution [3], and (or more generally) H-variate [19], [21] for various diffusion scenarios.…”

Section: A H-noise Modelmentioning

confidence: 99%

“…and its corresponding geometric power P (t) is given by P (t) = a 2 G [3]. Table IV shows the H-noise t and its geometric power P (t) for the typical anomalous diffusion models in Table III.…”

Section: Remark 10 (Normal Diffusion)mentioning

confidence: 99%

“…, y N } at the RN. Since H-noise can be the heavytailed random variable, it may have a large waiting time for molecules arriving at the RN, i.e., there exists a positive probability that the molecule will not have arrived at TN within finite 15 The number of molecule types used for this model can be minimized by introducing the lifetime of molecules [21], [62] and the chemical reactions in the medium [3], [63], [64]. ... time.…”

Section: A Molecular Communication System Modelmentioning

confidence: 99%

“…Nanotechnology has been highlighted for various applications such as medical systems, healthcare systems, nano-material, nano-machinary, nanoscale communication networks, and molecular communication systems. In particular, molecular communication is an emerging technology for communication between nanomachines where information is conveyed by means of molecules [1]- [3]. In passive transport molecular communication, the random propagation of molecules in a fluid medium such as air, water, or blood vessels in human tissue can be solely determined by the law of diffusion or can be subjected to unpredictable turbulence caused by the environment.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Molecular Communication in H-Diffusion

Trinh

Jeong

Shin

et al. 2020

IEEE Trans. Commun.

View full text Add to dashboard Cite

The random propagation of molecules in a fluid medium is characterized by the spontaneous diffusion law as well as the interaction between the environment and molecules. In this paper, we embody the anomalous diffusion theory for modeling and analysis in molecular communication. We employ H-diffusion to model a non-Fickian behavior of molecules in diffusive channels. H-diffusion enables us to model anomalous diffusion as the subordinate relationship between self-similar parent and directing processes and their corresponding probability density functions with two H-variates in a unified fashion. In addition, we introduce standard H-diffusion to make a bridge of normal diffusion across wellknown anomalous diffusions such as space-time fractional diffusion, Erdélyi-Kober fractional diffusion, grey Brownian motion, fractional Brownian motion, and Brownian motion. We then characterize the statistical properties of uncertainty of the random propagation time of a molecule governed by Hdiffusion laws by introducing a general class of molecular noise-called H-noise. Since H-noise can be an algebraic tailed process, we provide a concept of H-noise power using finite logarithm moments based on zero-order statistics. Finally, we develop a unifying framework for error probability analysis in a timing-based molecular communication system with a concept of signal-to-noise power ratio. ). 2 3 for timing and amplitude modulations. This work further extended to a connectivity problem with a random time constraint in a one-dimensional nanonetwork, where the random locations of molecules at the initial time are modeled by poisson point process [20]. The Cox process has been considered in [21] to capture the dynamic variation of the molecule concentration arising from the mobility of anomalously diffusive molecules, and the spatial ordering of the molecular communication performance has been characterized in terms of the error rate in the presence of interfering molecules. However, there is no comprehensive study on the modeling of anomalous diffusion channels in the context of molecular communication. Various anomalous diffusion processes typically can be modeled numerous ways including continuous random walk (CTRW), generalized diffusion equation, generalized master equation, fractional Brownian motion, and fractional kinetic equation (fractional diffusion equation) [15], [23]-[25]. 3 In particular, the CTRW simply describes diffusion of molecules in the medium with arbitrary distributions of jump lengths and waiting times. 4 In addition, the combination of a stochastic operational time-a directing process-and the self-similar parent process is equivalent to the subordination integral mechanism for the product of two random variables in the context of subordinated processes [28]-[31]. 5 This subordination law generates the solution of the fractional diffusion equation in purely analytical ways using the machinery offered by convolution properties of the Mellin transform. 6In this paper, we embody anomalous diffusion according to th...

show abstract