Transcript Prof. Suzuki slides "Simulator Development for Neuron

Simulator Development for
Neuron-based
Molecular Communication
Jun Suzuki
UMass Boston
Background
Nanoscale Communication
• An emerging research paradigm that aims to provide
communication capabilities between nanoscale
machines (nanomachines).
• The first publication at 2005
• Nanomachine
• The most basic functional unit in nanoscale systems.
• Scale: one to a few hundred nanometers.
• Consists of biological materials (e.g., molecules) and
perform very simple computation, sensing and/or
actuation tasks.
Nanomachines
Two Major Approaches to
Nanoscale Communication
• Electromagnetic communication
•
Top-down miniaturization approach
• Molecular communication
•
•
Bottom-up approach
Inspired by the communication mechanisms that naturally occur
among living cells.
Utilizes molecules as a communication medium.
Several advantages over electromagnetic communication
•
•
•
•
Inherent nanometer scale, biocompatibility and energy efficiency
A major application domain: in-body nanonetworks
•
Nanomachines are networked through molecular communication to
perform sensing and actuation tasks in the body for biomedical and
prosthetic purposes.
•
e.g., vital information sensing, targeted drug release and neural signal
augmentation.
Short- and Long-range
Molecular Communication
• Short-range communication
• nanometers to millimeters
• Uses molecular motors, calcium signaling and
bacteria communication
• Long-range communication
• millimeters to meters
• Uses neurons.
Neurons
•Fundamental components of the
nervous system.
– The central nervous system
• the brain, spinal cord and retina
– The peripheral nervous system
• sensory neurons, clusters of neurons
called ganglia
• Electrically excitable cells
• Process and transmit information by
electrical and chemical signaling.
The Structure of an Neuron
•Cell body (soma)
– 4 to 100 um in diameter
•Dendrites
•
•
•
•
•
Up to a few hundred um in length
Thin structures that arise from the soma
Form a complex “dendritic tree.”
the majority of inputs to a neuron
Generate the majority of inputs to a neuron.
Dendrite
Soma
Node of
ranvier
Axon terminal
Axon
Shwann cell
Nucleus
Myelin sheath
•An axon
•
•
•
•
Up to 1 m in humans
A special cellular extension that arises from the soma.
Branches hundreds, or even thousands, of times before it terminates.
Travels thru the body in bundles called nerves.
Dendrite
Soma
Node of
ranvier
Axon terminal
Axon
Shwann cell
Nucleus
Myelin sheath
Neuronal Network
•Neurons self-organize to connect each other and form a network.
Neuron-to-Neuron Communication
•Neurons communicate with others via synapses.
•A synapse
– A membrane-to-membrane junction
– Contains molecular machinery that allows a (presynaptic) neuron to transmit a
chemical signal to another (postsynaptic) neuron.
•An axon makes thousands of synaptic contacts.
Signal
Synapse
Presynaptic
Neuron
Postsynaptic
Neuron
Neurotransmitters
•When an axon terminal is
stimulated with an electric signal,
it initiates a chemical process in
a synapse.
– Generates neurotransmitter
molecules.
• e.g., Acetylcholine (ACh)
•Neurotransmitters electrically
excites a postsynaptic neuron.
Action Potential
•When a neuron is electrically excited, it generates an electrochemical pulse called
action potential.
•The refractory period follows a signal transmission.
•Signal transmission speed: 90 m (295 fts, 98 yds)/second
Neuron-based IBSANs
The Proposed Communication Framework
for Neuron-based IBSANs
•Neuronal TDMA
– TDMA (Time Division Multiple Access) protocol framework
TDMA Optimization
•Which neuron should start signaling in a given time slot?
– Assumption: Each neuron transmits at least one signal in the signaling cycle.
•Optimization objectives
– Signaling yield
• How many signals the sink receives from all sensors during the scheduling cycle.
– Signaling fairness among sensors
• How fairly/equally sensors access the shared neuronal network.
– Signaling delay
•A multiobjective optimization problem
Current Status
•A multiobjective genetic algorithm (GA) has been developed and
tested.
– Can optimize TDMA scheduling by balancing the trade-offs among 3
objectives.
What’s the Future of MolCom?
•Nanomachine development has been actively carried out in the academia and
industry.
•Molecular communication protocols/mechanisms have been studied through
simulations in the academia since mid ’00.
•Practical nanomachines will be developed in 10 (?) years.
•Molecular communication will be integrated with nanomachines.
•Applications
– Metical monitoring
– Prosthetic and rehabilitation devices
Cellular System Development and
TDMA
•TDMA
– is/was used in 2G cellular systems such as GSM.
– is still widely used in satellite systems.
– was extensively studied in the academia for wireless communication in 70s and early 80s
•through simulations.
•GSM
–
–
–
–
The standardization of GSM started in ’82 and completed in ’90.
The first GSM call in 1991.
Practically usable cell phones became available on the market around 1993 - 1995.
GSM subscribers worldwide passed 10M in 1995, 100M in 1998, 500M in 2001, 1B in 2004
and 2B in 2008.
– Has been fading out as 3G and 4G systems emerge.
CS682/3 Project:
A Generic Simulation Environment for
Neuronal TDMA
Architectural Overview
–Neuronal Network Visualizer/Editor
• Aids the user to manually draw neuronal networks in 2D/3D
• Performs several visualization effects
• e.g., zoom in, zoom out, etc.
• Generates/serializes a text file (in XML?) according to a drawn neuronal network and stores it in the
local disk or a remote cloud (e.g., Dropbox and Google Drive).
• Reads a text file and visualize a neuronal network.
Topology Data
Generator
GUI
Frontend
Topology Data
Reader
Topology
Data
(XML?)
Manual 2D/3D
topology editor
Automatic
Topology
Generator
Optimization
Code
Opt. Algo.
(NSGA-II)
Opt. Engine
(jMetal)
Optimization
Code
Opt. Algo.
(Simplex)
Opt. Engine
(GLPK)
–Automatic Topology Generator
• Artificially grow neuronal networks based on a topology construction algorithm(s) and serializes
their topology information in text files (XML?).
•Optimization Code
• I/F code to run optimization algorithms with a given neuronal network topology.
Topology Data
Generator
GUI
Frontend
Topology Data
Reader
Topology
Data
(XML?)
Manual 2D/3D
topology editor
Automatic
Topology
Generator
Optimization
Code
Opt. Algo.
(NSGA-II)
Opt. Engine
(jMetal)
Optimization
Code
Opt. Algo.
(Simplex)
Opt. Engine
(GLPK)
What You can Gain from this Project.
•Extensive Java programming
– GUI, 2D/3D visualization, text data handling, optimization, self-organizing
topology formation
•Very futuristic research experience
– Potential co-authored publications (if you are interested)