Biology-informed communication protocols for bio-molecular networks

Wang, Jiaming

Permalink

https://hdl.handle.net/2142/127419 Copy

Description

Title

Biology-informed communication protocols for bio-molecular networks

Author(s)

Wang, Jiaming

Issue Date

2024-11-26

Director of Research (if dissertation) or Advisor (if thesis)

Al-Hassanieh, Haitham

Doctoral Committee Chair(s)

Al-Hassanieh, Haitham

Committee Member(s)

• Roy Choudhury, Romit
• Srikant, Rayadurgam
• Krishnaswamy, Bhuvana

Department of Study

Electrical & Computer Eng

Discipline

Electrical & Computer Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Molecular Communication
• Wireless Communication
• Internet of Bio-Nano Things
• Body Area Network.

Abstract

The field of synthetic biology and bioengineering has made significant strides in the development of microscale and nano-scale bio-implants that can operate inside the human body, which has led to an expansion of research interest in developing new networks for these in-body devices. Due to the small factor and bio-compatibility requirements of bio-implants, traditional wireless technologies are unsuitable. Instead, molecular communication (MC) has emerged as a promising alternative to fill this gap. Molecular communication is a scheme that uses particles as information carriers to exchange messages between devices, which can be easily implemented by manipulating the release of these information-bearing particles–the most straightforward method is using the presence or absence of these molecules to deliver “0” or “1” bits. Molecular communication stems mainly from chemistry and biology, making it bio-compatible and highly energy-efficient for operation with bio-implants inside the human body. While molecular communication has been studied in synthetic biology and information theory, there is little research on building molecular communication systems that account for the unique challenges of communication inside the human body. This thesis aims to design and implement efficient biology-informed bio-molecular communication protocols that account for the unique challenges of in-body networks. We start by building an understanding of the bio-molecular communication channel and study how the molecular signals propagate inside the blood vessels. We highlight new properties of the MC channel that differ from traditional communication channels and that were not accounted for by past work. By understanding the biological properties and constraints, we identify the challenges and opportunities to bridge the gap between theory and practice. In particular, our design is based on experimental setups that emulate the blood flow inside the body. Leveraging experimental setups allows us to highlight the following challenges that fall into three categories: • Challenges for general molecular channels. Based on the theory of particle propagation under diffusion and advection, two known challenges for molecular communication: (1) Long delay spread. The signals released by the transmitter at the same time can arrive at the receiver through different propagation delays. Such spreading is not only longer than the expected propagation delay, but also positively related to it. This leads to the inter-symbol interference (ISI) in communication. (2) Signal-dependent noise. Due to diffusion and unpredictable flows, the propagation of information-bearing particles is highly random. Thus, the number of particles detected by the receiver follows a probability distribution that the variance (channel noise), instead of some constant, is highly related to the expectation (channel response). This dissertation not only verifies the above two challenges in a blood-vessel-like testbed but also observes two other general challenges in a molecular channel: (1) Non-causal channel. The randomness in the propagation delay indicates that a non-negligible proportion of particles will arrive earlier than expected. Moreover, particles released in one symbol can arrive earlier than the majority of particles from previous symbols, which is more prominent under high transmission rate. (2) Short coherence time. The variation of molecular channels is around tens of seconds. Although this number is significantly larger than conventional wireless channels, the ratio to the delay spread is however much smaller, which highly restricts the efficiency of data packets with conventional techniques. • Challenges for multiple access. This dissertation also highlights two challenges that cause trouble in designing fair and scalable communication protocols among multiple transmitters when interference from other devices comes into play. (1) Lack of synchronization. Particles are more inclined to propagate along the direction of environmental flow. Because of this, the molecular channel is asymmetric in that the downstream device can receive the signal from upstream, but not the opposite. This unique property decides that most conventional multiple access protocols are either infeasible (if they require synchronization or feedback) or compromised in fairness and scalability (such as carrier sensing). (2) Non-negative signal. Molecular signals are some representations of the number of information-bearing particles. Thus, in a molecular network with a single type of particle, the signals are non-negative, which cannot be cancelled like wireless signals with phases. Multi-User Interference (MUI) remains a problem in molecular channels. • Challenges for blood flow. The unique phenomenon in the blood vessel also raises a challenge for communication, while no similar issues can be found in conventional wireless systems: Heartbeat-induced varying channel. Heartbeat leads to the periodic variation of the blood flow in the vessels, especially in the arteries that are close to the heart where the maximum flow can be 3× more than the minimum. The normal heartbeat period has roughly the same order of magnitude as the packet symbol interval, which leads to the evident channel variation across every symbol. To address these biology-informed communication challenges, this dissertation proposes new protocols for each of the categories, which are not contradictory and can be stacked up together as a whole: • μ-Link for general molecular channel. Since the received signal for each symbol in a molecular packet is highly influenced by the interference from other symbols, μ-Link uses Hidden Markov Model (HMM) to model the molecular channel as the basic solution to track the long ISI–both causal and non-causal. Viterbi algorithm is adopted to solve the problem, while a channel tracking module is added to periodically compensate the divergence of the channel response beyond the coherence time. μ-Link is able to achieve comparable performance to a well-trained deep learning solution. • MoMA for multiple access. Because of asymmetry, protocols that rely on feedback or sensing are not suitable for molecular channels. Taking this into consideration, this dissertation extends the CDMA (Code-Division Multiple Access) and proposes MoMA (Molecular Multiple Access). Since collision is inevitable in molecular network, MoMA carefully chooses the codebook, symbol representation and packet construction to facilitate the detection and decoding of collided packets on the receiver end. Empirical knowledge of molecular channels is introduced to further reduce the error and extend scalability. MoMA is able to achieve a 70% higher throughput than existing CDMA solutions in a network of four transmitters with frequent collisions. • FlowLink for blood flow. In order to track the channel variation per symbol, we need to stick to the invariable in the molecular channel. FlowLink’s core idea is an observation that the total amount of flow required for the propagation, i.e. the integration of the flow rate over the propagation period, is a constant. Instead of modeling the molecular signal as a function of time, it models the signal as a function of flow. By transforming the molecular signal from time samples to flow samples, FlowLink unifies the channel response in the new representation, which can then be fed into a modified decoder with respect to flow interval. FlowLink demonstrates obvious improvement on more than 90% of molecular packets, while shows little negative influence on the rest data. This dissertation also proposes an in-vitro (outside a living organism) testbed over which the proposed protocols are evaluated. It simulates part of the human circulatory system, which holds similar properties as the blood vessels, including material, diameter, topology, and blood flow. The dataset comprises extensive samples over varying data rate, TX-RX distance, flow cycle and other parameters that will help the illustration of the contributions.

Graduation Semester

2024-12

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/127419

Copyright and License Information

Copyright 2024 Jiaming Wang

Owning Collections