Model-Based: End-to-End Molecular Communication System Through Deep Reinforcement Learning Auto Encoder

Mohamed, Soha; Dong, Jian; Junejo, A. R.; Zuo, De Cheng

doi:10.1109/access.2019.2916701

Cited by 37 publications

(16 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While the previous studies merely focus on the detector design at the receiver side, the transmitter can also be considered for the joint transceiver optimization. In this case, the transmitter and receiver can be trained by the deep reinforcement learning and supervised learning, respectively [14]. Finally, the results based on the joint optimization display a better BER performance than the model-based threshold detection.…”

Section: B Comparison With Model-based Schemesmentioning

confidence: 99%

Signal Detection for Molecular Communication: Model-Based vs. Data-Driven Methods

Huang

Wei

et al. 2021

IEEE Commun. Mag.

View full text Add to dashboard Cite

Multi-scale molecular communication (MC) employs the characteristics of information molecules for information exchange. The received signal in MC inevitably encounters severe inter-symbol interference and signal-dependent noise due to the stochastic diffusion mechanism. Focusing on the critical signal detection in MC, first this article reviews the commonly used model-based detectors, and exposes their limitations in practical implementation. Then, the emerging data-driven detectors that can make up for some deficiencies of the model-based detectors are presented. Despite the black-box nature of the data-driven detectors, the explainable artificial intelligence can be further investigated for the performance improvement of transparency and trust. I. INTRODUCTION Molecular communication (MC) is an emerging multi-scale communication paradigm that conveys messages via the patterns of molecules [1]. MC has been regarded as an alternative for conventional communication techniques in virtue of its unique transmission mechanism.For the nano-scale and micro-scale channels in vivo, the implementation of the conventional communication techniques is hindered for the following reasons. The electromagnetic (EM) wave based wireless communication has its constraint in wave generation. As the wavelength in operation is proportional to the antenna size, nanomachine requires the Tera-hertz band for communications, which is, however, still in its research infancy, let alone the micro-scale implementation. Besides, optical communication is not an appropriate choice due to complex cell/tissue obstacles in the channel. Finally, acoustic communication has an energy issue on such an extremely tiny scale. In addition to the implementation restrictions, the bio-compatibility issue of the above schemes has not been verified in the open literature. As for the macro-scale channels, the EM wave based communication has the propagation problem in metal and aqueous environments. Inspired by the biological systems, MC can remedy the flaws in the existing communication systems. For micro-scale channels, MC via diffusion (MCvD) is an energy-efficient signal transmission mechanism that uses information molecules from the communication environments, and no extra energy is required for the

show abstract

Section: B Comparison With Model-based Schemesmentioning

confidence: 99%

Signal Detection for Molecular Communication: Model-Based vs. Data-Driven Methods

Huang

Wei

et al. 2021

IEEE Commun. Mag.

View full text Add to dashboard Cite

show abstract

“…This approach has also been used successfully in OFDM systems [43] to learn the symbols transmitted over the subcarriers. Deep learning techniques and auto-encoder style training have been used in the fields of fiber-optic [44], [45] and molecular communication [46], [47] to model the channel and to leverage the channel model to learn communication schemes that achieve low error rates.…”

Section: Related Workmentioning

confidence: 99%

Blind Interactive Learning of Modulation Schemes: Multi-Agent Cooperation Without Co-Design

et al. 2020

View full text Add to dashboard Cite

We examine the problem of learning to cooperate in the context of wireless communication. In our setting, two agents must learn modulation schemes that enable them to communicate across a power-constrained additive white Gaussian noise channel. We investigate whether learning is possible under different levels of information sharing between distributed agents which are not necessarily co-designed. We employ the ''Echo'' protocol, a ''blind'' interactive learning protocol where an agent hears, understands, and repeats (echoes) back the message received from another agent, simultaneously training itself to communicate. To capture the idea of cooperation between ''not necessarily co-designed'' agents we use two different populations of function approximators-neural networks and polynomials. We also include interactions between learning agents and non-learning agents with fixed modulation protocols such as QPSK and 16QAM. We verify the universality of the Echo learning approach, showing it succeeds independent of the inner workings of the agents. In addition to matching the communication expectations of others, we show that two learning agents can collaboratively invent a successful communication approach from independent random initializations. We complement our simulations with an implementation of the Echo protocol in software-defined radios. To explore the continuum of co-design, we study how learning is impacted by different levels of information sharing between agents, including sharing training symbols, losses, and full gradients. We find that co-design (increased information sharing) accelerates learning. Learning higher order modulation schemes is a more difficult task, and the beneficial effect of co-design becomes more pronounced as the task becomes harder.

show abstract

“…and describes the input/output relation of the extrinsic mutual information of the iterative demapper [17], i.e., for a given a priori mutual information, how much (additional) extrinsic information the demapper can contribute by observing the channel output. This EXIT characteristic is defined 4 as…”

Section: A Exit Analysismentioning

confidence: 99%

“…This approach enables joint optimization of the transmitter and receiver for a specific channel model without extensive mathematical analysis. Autoencoder-based communication systems have first been proposed in the context of wireless communications [1], and have subsequently been extended towards other settings, such as optical fiber [2], optical wireless [3], and molecular communications [4]. Most of these approaches are optimized on the symbol-wise categorical cross entropy (CE), which is equivalent to maximizing the mutual information between the channel input and output [5].…”

Section: Introductionmentioning

confidence: 99%

Trainable Communication Systems: Concepts and Prototype

Cammerer¹,

Aoudia²,

Dörner³

et al. 2019

Preprint

View full text Add to dashboard Cite

We consider a trainable point-to-point communication system, where both transmitter and receiver are implemented as neural networks (NNs), and demonstrate that training on the bit-wise mutual information (BMI) allows seamless integration with practical bit-metric decoding (BMD) receivers, as well as joint optimization of constellation shaping and labeling. Moreover, we present a fully differentiable neural iterative demapping and decoding (IDD) structure which achieves significant gains on additive white Gaussian noise (AWGN) channels using a standard 802.11n low-density parity-check (LDPC) code. The strength of this approach is that it can be applied to arbitrary channels without any modifications. Going one step further, we show that careful code design can lead to further performance improvements. Lastly, we show the viability of the proposed system through implementation on software-defined radios (SDRs) and training of the end-to-end system on the actual wireless channel. Experimental results reveal that the proposed method enables significant gains compared to conventional techniques.

show abstract

Model-Based: End-to-End Molecular Communication System Through Deep Reinforcement Learning Auto Encoder

Cited by 37 publications

References 28 publications

Signal Detection for Molecular Communication: Model-Based vs. Data-Driven Methods

Signal Detection for Molecular Communication: Model-Based vs. Data-Driven Methods

Blind Interactive Learning of Modulation Schemes: Multi-Agent Cooperation Without Co-Design

Trainable Communication Systems: Concepts and Prototype

Contact Info

Product

Resources

About