Biological Systems for Intrusion Detection
Content
Biological Programming Models for IntrusionTolerant Systems
Workshop on Statistical and Machine Learning Techniques in Computer Intrusion Detection George Mason University 24 September 2003
David Evans
University of Virginia,
Department of Computer Science
Learning from Biology
• Process
– Genetic algorithms
• Product
– Immune systems – Programs
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
2
(Really) Brief History of Computing
1950 1960 1970 1980 1990 2000Billions of small, cheap unreliable devices Computing organized through local interactions Fundamentally integrated into physical environment
3
Monolithic Computers in guarded, airconditioned rooms No interactions
Fixed Networks of PCs
Data interactions with other computers, but most computing done locally Rich interface to user, limited interface to environment
swarm.cs.virginia.edu
Narrow interface to operator (punch cards, teletype), no interface to environment
ID03 - 24 Sept 2003
Challenges and Opportunities
• Embedded in physical environment
– Challenges: unpredictable, energy-limited – Opportunities: physical laws, continuous
• Scale
– Challenges: billions of independent components – Opportunities: redundant to failures
• Demands new programming approaches and reasoning techniques
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
4
Swarm Computing: Long-Range Goal
Cement 10 TFlop
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
5
Existing Systems
• Trillions of unreliable, inexpensive components • No significant performance degradation when billions fail • Small programs produce complex behavior
ID03 - 24 Sept 2003
• ~70 Trillion Cells in adult human • ~ 3 Billion of your cells have died since I started talking • Program (3B nucleotides) is shorter than Windows XP, difference between 2 humans on 1 floppy
6
swarm.cs.virginia.edu
Swarm Programming
Behavioral Description Environment Model Device Model Behavior and primitives defined over groups Device Units
Swarm Program Generator
Primitives Library
Device Programs
Programmed Device Units
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
7
Observations About Nature’s Programs
• Responsive
– Aware of state of self and surroundings
• Localized
– Communication through chemical diffusion
• Redundant
– Millions of cells can die without compromising function
• Diverse
– Species survive because of diversity of individuals
• Remarkably Expressive
• Human genome ~250MB
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
8
Foundations
Current Research
• Amorphous Computing [Abelson, Nagpal, Sussman] Cellular Automata • Paintable Computing von Neumann [1940s] [Butera] Conway’s Game of Life [1970] • Embryonics [Mange, Wolfram [2002] Sipper] Reaction-Diffusion Turing [1952] • Ant Colony Optimization, Swarm Intelligence
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
9
Simplified Cell Model
• Awareness of Environment
– Sense chemicals on cell walls – Sense chemicals in environment
• Cell Actions
– Cell Division (asymmetric) – State Change – Communicate: emit (directional, neighboring walls), diffuse (omnidirectional)
• Simple physical forces
– Two cells cannot overlap in space
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
10
Biological Complexity
Molecular map of colon cancer cell from http://www.gnsbiotech.com/applications.shtml
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
11
Simple Sphere Program
center
alive < 1
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions alive from dir < 1 -> (center, body) in dir; }
body
state body { color 0 0 1 emits (alive, 1) transitions alive from dir < 1 & radius > 0 alive < 1 & radius > 0 -> (body, body) in dir; }
swarm.cs.virginia.edu
ID03 - 24 Sept 2003
12
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions alive from dir < 1 -> (center, body) in dir; } state body { color 0 0 1 emits (alive, 1) transitions alive from dir < 1 & radius > 0 -> (body, body) in dir; }
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
13
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
14
Robustness of Sphere Program
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
15
Intrusion Tolerance?
• Robust to random failures
– As long as source cell survives, the sphere will re-generate – Sphere has > 10000 cells
• Not robust to attacks
– Destroy the center cell, sphere will not regrow
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
16
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions (alive from dir < 1) -> (center, core) in dir; }
Example
state core { color 0 1 0 emits (alive, 1) transitions (alive from dir < 1) & (radius > 2) -> (core, body) in dir; (radius < 2) & (alive from dir < 1) -> (core, center) in dir; } state body { color 1 1 0 emits (alive, 1) transitions (alive from dir < 1) & (radius > 1) -> (body, body) in dir; }
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
17
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
18
state corner { color red emits (length, 8), (alive, 1) transitions (alive < 1) from dir -> (corner, segment) in dir; -> (corner); } state segment { color cyan emits (alive, 1) forwards (length - 1) transitions (length > 1.5) from dir & (alive < 0.5) from opposite (dir) -> (segment, segment) in opposite (dir); (length > 0.1) -> (corner); (length < 0.1) -> die; }
ID03 - 24 Sept 2003
Network Mesh
swarm.cs.virginia.edu
19
Composing Primitives
• Cells can follow multiple programs simultaneously (vector of independent states) • Cells can combine primitives through shared chemicals
– Chemicals secreted by one primitive can induce changes in other primitives
• Goals:
– Predict properties of composition based on properties of primitives – Diversity of primitive implementations provides protection from directed attacks
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
20
Mickey Mouse Program • 20 states • 50 transition rules • Starts from one cell, combines lines, spheres Real Mouse Program • 3B base pairs • 98% same as human DNA • Starts from one cell, combines complex proteins
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
21
Towards Real Systems
• Cells
– Sensor Devices, MEMS, Internet Nodes
• Division
– Processes – Find new hosts
• Communication
– Point-to-point emissions – Wireless multicast (can be multi-hop) diffusions
• Example: distributed file system running on simulated wireless nodes
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
22
Distributed Wireless File Service
File Distribution and Update
Publishe r
inhibit
publish
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
23
Distributed Wireless File Service
File Distribution and Update
replicate
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
24
Robustness to Node Failures
80% of requests satisfied
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
25
Summary
• Trillions of creatures have died to evolve the extremely robust programs that survive today • Robustness and scalability require:
– Decentralization – Awareness of surroundings – Diversity
• Swarm Programming
– Develop high level behaviors from local interactions – Use communication through environment to coordinate locally – Produce complex behaviors by combining primitives defined over groups
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
26
Acknowledgements
Sponsor: National Science Foundation Contributors: Lance Davidson (UVa Biology) Selvin George Salvatore Guarnieri Steven Marchette Qi Wang Brian Zhang
http://swarm.cs.virginia.edu
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
27
Biological Programming Models for IntrusionTolerant Systems
Workshop on Statistical and Machine Learning Techniques in Computer Intrusion Detection George Mason University 24 September 2003 ID03 - 24 Sept 2003
David Evans
University of Virginia,
Department of Computer Science
1
swarm.cs.virginia.edu
Learning from Biology
• Process
– Genetic algorithms
• Product
– Immune systems – Programs
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
2
(Really) Brief History of Computing
1950 1960 1970 1980 1990 2000Billions of small, cheap unreliable devices Computing organized through local interactions Fundamentally integrated into physical environment
3
Monolithic Computers in guarded, airconditioned rooms No interactions
Fixed Networks of PCs
Data interactions with other computers, but most computing done locally Rich interface to user, limited interface to environment
swarm.cs.virginia.edu
Narrow interface to operator (punch cards, teletype), no interface to environment
ID03 - 24 Sept 2003
Challenges and Opportunities
• Embedded in physical environment
– Challenges: unpredictable, energy-limited – Opportunities: physical laws, continuous
• Scale
– Challenges: billions of independent components – Opportunities: redundant to failures
• Demands new programming approaches and reasoning techniques
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
4
Swarm Computing: Long-Range Goal
Cement 10 TFlop
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
5
Existing Systems
• Trillions of unreliable, inexpensive components • No significant performance degradation when billions fail • Small programs produce complex behavior
ID03 - 24 Sept 2003
• ~70 Trillion Cells in adult human • ~ 3 Billion of your cells have died since I started talking • Program (3B nucleotides) is shorter than Windows XP, difference between 2 humans on 1 floppy
6
swarm.cs.virginia.edu
Swarm Programming
Behavioral Description Environment Model Device Model Behavior and primitives defined over groups Device Units
Swarm Program Generator
Primitives Library
Device Programs
Programmed Device Units
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
7
Observations About Nature’s Programs
• Responsive
– Aware of state of self and surroundings
• Localized
– Communication through chemical diffusion
• Redundant
– Millions of cells can die without compromising function
• Diverse
– Species survive because of diversity of individuals
• Remarkably Expressive
• Human genome ~250MB
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
8
Foundations
Current Research
• Amorphous Computing [Abelson, Nagpal, Sussman] Cellular Automata • Paintable Computing von Neumann [1940s] [Butera] Conway’s Game of Life [1970] • Embryonics [Mange, Wolfram [2002] Sipper] Reaction-Diffusion Turing [1952] • Ant Colony Optimization, Swarm Intelligence
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
9
Simplified Cell Model
• Awareness of Environment
– Sense chemicals on cell walls – Sense chemicals in environment
• Cell Actions
– Cell Division (asymmetric) – State Change – Communicate: emit (directional, neighboring walls), diffuse (omnidirectional)
• Simple physical forces
– Two cells cannot overlap in space
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
10
Biological Complexity
Molecular map of colon cancer cell from http://www.gnsbiotech.com/applications.shtml
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
11
Simple Sphere Program
center
alive < 1
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions alive from dir < 1 -> (center, body) in dir; }
body
state body { color 0 0 1 emits (alive, 1) transitions alive from dir < 1 & radius > 0 alive < 1 & radius > 0 -> (body, body) in dir; }
swarm.cs.virginia.edu
ID03 - 24 Sept 2003
12
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions alive from dir < 1 -> (center, body) in dir; } state body { color 0 0 1 emits (alive, 1) transitions alive from dir < 1 & radius > 0 -> (body, body) in dir; }
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
13
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
14
Robustness of Sphere Program
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
15
Intrusion Tolerance?
• Robust to random failures
– As long as source cell survives, the sphere will re-generate – Sphere has > 10000 cells
• Not robust to attacks
– Destroy the center cell, sphere will not regrow
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
16
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions (alive from dir < 1) -> (center, core) in dir; }
Example
state core { color 0 1 0 emits (alive, 1) transitions (alive from dir < 1) & (radius > 2) -> (core, body) in dir; (radius < 2) & (alive from dir < 1) -> (core, center) in dir; } state body { color 1 1 0 emits (alive, 1) transitions (alive from dir < 1) & (radius > 1) -> (body, body) in dir; }
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
17
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
18
state corner { color red emits (length, 8), (alive, 1) transitions (alive < 1) from dir -> (corner, segment) in dir; -> (corner); } state segment { color cyan emits (alive, 1) forwards (length - 1) transitions (length > 1.5) from dir & (alive < 0.5) from opposite (dir) -> (segment, segment) in opposite (dir); (length > 0.1) -> (corner); (length < 0.1) -> die; }
ID03 - 24 Sept 2003
Network Mesh
swarm.cs.virginia.edu
19
Composing Primitives
• Cells can follow multiple programs simultaneously (vector of independent states) • Cells can combine primitives through shared chemicals
– Chemicals secreted by one primitive can induce changes in other primitives
• Goals:
– Predict properties of composition based on properties of primitives – Diversity of primitive implementations provides protection from directed attacks
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
20
Mickey Mouse Program • 20 states • 50 transition rules • Starts from one cell, combines lines, spheres Real Mouse Program • 3B base pairs • 98% same as human DNA • Starts from one cell, combines complex proteins
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
21
Towards Real Systems
• Cells
– Sensor Devices, MEMS, Internet Nodes
• Division
– Processes – Find new hosts
• Communication
– Point-to-point emissions – Wireless multicast (can be multi-hop) diffusions
• Example: distributed file system running on simulated wireless nodes
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
22
Distributed Wireless File Service
File Distribution and Update
Publishe r
inhibit
publish
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
23
Distributed Wireless File Service
File Distribution and Update
replicate
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
24
Robustness to Node Failures
80% of requests satisfied
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
25
Summary
• Trillions of creatures have died to evolve the extremely robust programs that survive today • Robustness and scalability require:
– Decentralization – Awareness of surroundings – Diversity
• Swarm Programming
– Develop high level behaviors from local interactions – Use communication through environment to coordinate locally – Produce complex behaviors by combining primitives defined over groups
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
26
Acknowledgements
Sponsor: National Science Foundation Contributors: Lance Davidson (UVa Biology) Selvin George Salvatore Guarnieri Steven Marchette Qi Wang Brian Zhang
http://swarm.cs.virginia.edu
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
27
Workshop on Statistical and Machine Learning Techniques in Computer Intrusion Detection George Mason University 24 September 2003
David Evans
University of Virginia,
Department of Computer Science
Learning from Biology
• Process
– Genetic algorithms
• Product
– Immune systems – Programs
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
2
(Really) Brief History of Computing
1950 1960 1970 1980 1990 2000Billions of small, cheap unreliable devices Computing organized through local interactions Fundamentally integrated into physical environment
3
Monolithic Computers in guarded, airconditioned rooms No interactions
Fixed Networks of PCs
Data interactions with other computers, but most computing done locally Rich interface to user, limited interface to environment
swarm.cs.virginia.edu
Narrow interface to operator (punch cards, teletype), no interface to environment
ID03 - 24 Sept 2003
Challenges and Opportunities
• Embedded in physical environment
– Challenges: unpredictable, energy-limited – Opportunities: physical laws, continuous
• Scale
– Challenges: billions of independent components – Opportunities: redundant to failures
• Demands new programming approaches and reasoning techniques
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
4
Swarm Computing: Long-Range Goal
Cement 10 TFlop
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
5
Existing Systems
• Trillions of unreliable, inexpensive components • No significant performance degradation when billions fail • Small programs produce complex behavior
ID03 - 24 Sept 2003
• ~70 Trillion Cells in adult human • ~ 3 Billion of your cells have died since I started talking • Program (3B nucleotides) is shorter than Windows XP, difference between 2 humans on 1 floppy
6
swarm.cs.virginia.edu
Swarm Programming
Behavioral Description Environment Model Device Model Behavior and primitives defined over groups Device Units
Swarm Program Generator
Primitives Library
Device Programs
Programmed Device Units
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
7
Observations About Nature’s Programs
• Responsive
– Aware of state of self and surroundings
• Localized
– Communication through chemical diffusion
• Redundant
– Millions of cells can die without compromising function
• Diverse
– Species survive because of diversity of individuals
• Remarkably Expressive
• Human genome ~250MB
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
8
Foundations
Current Research
• Amorphous Computing [Abelson, Nagpal, Sussman] Cellular Automata • Paintable Computing von Neumann [1940s] [Butera] Conway’s Game of Life [1970] • Embryonics [Mange, Wolfram [2002] Sipper] Reaction-Diffusion Turing [1952] • Ant Colony Optimization, Swarm Intelligence
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
9
Simplified Cell Model
• Awareness of Environment
– Sense chemicals on cell walls – Sense chemicals in environment
• Cell Actions
– Cell Division (asymmetric) – State Change – Communicate: emit (directional, neighboring walls), diffuse (omnidirectional)
• Simple physical forces
– Two cells cannot overlap in space
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
10
Biological Complexity
Molecular map of colon cancer cell from http://www.gnsbiotech.com/applications.shtml
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
11
Simple Sphere Program
center
alive < 1
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions alive from dir < 1 -> (center, body) in dir; }
body
state body { color 0 0 1 emits (alive, 1) transitions alive from dir < 1 & radius > 0 alive < 1 & radius > 0 -> (body, body) in dir; }
swarm.cs.virginia.edu
ID03 - 24 Sept 2003
12
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions alive from dir < 1 -> (center, body) in dir; } state body { color 0 0 1 emits (alive, 1) transitions alive from dir < 1 & radius > 0 -> (body, body) in dir; }
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
13
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
14
Robustness of Sphere Program
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
15
Intrusion Tolerance?
• Robust to random failures
– As long as source cell survives, the sphere will re-generate – Sphere has > 10000 cells
• Not robust to attacks
– Destroy the center cell, sphere will not regrow
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
16
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions (alive from dir < 1) -> (center, core) in dir; }
Example
state core { color 0 1 0 emits (alive, 1) transitions (alive from dir < 1) & (radius > 2) -> (core, body) in dir; (radius < 2) & (alive from dir < 1) -> (core, center) in dir; } state body { color 1 1 0 emits (alive, 1) transitions (alive from dir < 1) & (radius > 1) -> (body, body) in dir; }
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
17
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
18
state corner { color red emits (length, 8), (alive, 1) transitions (alive < 1) from dir -> (corner, segment) in dir; -> (corner); } state segment { color cyan emits (alive, 1) forwards (length - 1) transitions (length > 1.5) from dir & (alive < 0.5) from opposite (dir) -> (segment, segment) in opposite (dir); (length > 0.1) -> (corner); (length < 0.1) -> die; }
ID03 - 24 Sept 2003
Network Mesh
swarm.cs.virginia.edu
19
Composing Primitives
• Cells can follow multiple programs simultaneously (vector of independent states) • Cells can combine primitives through shared chemicals
– Chemicals secreted by one primitive can induce changes in other primitives
• Goals:
– Predict properties of composition based on properties of primitives – Diversity of primitive implementations provides protection from directed attacks
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
20
Mickey Mouse Program • 20 states • 50 transition rules • Starts from one cell, combines lines, spheres Real Mouse Program • 3B base pairs • 98% same as human DNA • Starts from one cell, combines complex proteins
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
21
Towards Real Systems
• Cells
– Sensor Devices, MEMS, Internet Nodes
• Division
– Processes – Find new hosts
• Communication
– Point-to-point emissions – Wireless multicast (can be multi-hop) diffusions
• Example: distributed file system running on simulated wireless nodes
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
22
Distributed Wireless File Service
File Distribution and Update
Publishe r
inhibit
publish
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
23
Distributed Wireless File Service
File Distribution and Update
replicate
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
24
Robustness to Node Failures
80% of requests satisfied
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
25
Summary
• Trillions of creatures have died to evolve the extremely robust programs that survive today • Robustness and scalability require:
– Decentralization – Awareness of surroundings – Diversity
• Swarm Programming
– Develop high level behaviors from local interactions – Use communication through environment to coordinate locally – Produce complex behaviors by combining primitives defined over groups
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
26
Acknowledgements
Sponsor: National Science Foundation Contributors: Lance Davidson (UVa Biology) Selvin George Salvatore Guarnieri Steven Marchette Qi Wang Brian Zhang
http://swarm.cs.virginia.edu
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
27
Biological Programming Models for IntrusionTolerant Systems
Workshop on Statistical and Machine Learning Techniques in Computer Intrusion Detection George Mason University 24 September 2003 ID03 - 24 Sept 2003
David Evans
University of Virginia,
Department of Computer Science
1
swarm.cs.virginia.edu
Learning from Biology
• Process
– Genetic algorithms
• Product
– Immune systems – Programs
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
2
(Really) Brief History of Computing
1950 1960 1970 1980 1990 2000Billions of small, cheap unreliable devices Computing organized through local interactions Fundamentally integrated into physical environment
3
Monolithic Computers in guarded, airconditioned rooms No interactions
Fixed Networks of PCs
Data interactions with other computers, but most computing done locally Rich interface to user, limited interface to environment
swarm.cs.virginia.edu
Narrow interface to operator (punch cards, teletype), no interface to environment
ID03 - 24 Sept 2003
Challenges and Opportunities
• Embedded in physical environment
– Challenges: unpredictable, energy-limited – Opportunities: physical laws, continuous
• Scale
– Challenges: billions of independent components – Opportunities: redundant to failures
• Demands new programming approaches and reasoning techniques
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
4
Swarm Computing: Long-Range Goal
Cement 10 TFlop
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
5
Existing Systems
• Trillions of unreliable, inexpensive components • No significant performance degradation when billions fail • Small programs produce complex behavior
ID03 - 24 Sept 2003
• ~70 Trillion Cells in adult human • ~ 3 Billion of your cells have died since I started talking • Program (3B nucleotides) is shorter than Windows XP, difference between 2 humans on 1 floppy
6
swarm.cs.virginia.edu
Swarm Programming
Behavioral Description Environment Model Device Model Behavior and primitives defined over groups Device Units
Swarm Program Generator
Primitives Library
Device Programs
Programmed Device Units
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
7
Observations About Nature’s Programs
• Responsive
– Aware of state of self and surroundings
• Localized
– Communication through chemical diffusion
• Redundant
– Millions of cells can die without compromising function
• Diverse
– Species survive because of diversity of individuals
• Remarkably Expressive
• Human genome ~250MB
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
8
Foundations
Current Research
• Amorphous Computing [Abelson, Nagpal, Sussman] Cellular Automata • Paintable Computing von Neumann [1940s] [Butera] Conway’s Game of Life [1970] • Embryonics [Mange, Wolfram [2002] Sipper] Reaction-Diffusion Turing [1952] • Ant Colony Optimization, Swarm Intelligence
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
9
Simplified Cell Model
• Awareness of Environment
– Sense chemicals on cell walls – Sense chemicals in environment
• Cell Actions
– Cell Division (asymmetric) – State Change – Communicate: emit (directional, neighboring walls), diffuse (omnidirectional)
• Simple physical forces
– Two cells cannot overlap in space
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
10
Biological Complexity
Molecular map of colon cancer cell from http://www.gnsbiotech.com/applications.shtml
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
11
Simple Sphere Program
center
alive < 1
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions alive from dir < 1 -> (center, body) in dir; }
body
state body { color 0 0 1 emits (alive, 1) transitions alive from dir < 1 & radius > 0 alive < 1 & radius > 0 -> (body, body) in dir; }
swarm.cs.virginia.edu
ID03 - 24 Sept 2003
12
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions alive from dir < 1 -> (center, body) in dir; } state body { color 0 0 1 emits (alive, 1) transitions alive from dir < 1 & radius > 0 -> (body, body) in dir; }
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
13
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
14
Robustness of Sphere Program
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
15
Intrusion Tolerance?
• Robust to random failures
– As long as source cell survives, the sphere will re-generate – Sphere has > 10000 cells
• Not robust to attacks
– Destroy the center cell, sphere will not regrow
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
16
state center { color 1 0 0 emits (alive, 1) diffuses (radius, 10) transitions (alive from dir < 1) -> (center, core) in dir; }
Example
state core { color 0 1 0 emits (alive, 1) transitions (alive from dir < 1) & (radius > 2) -> (core, body) in dir; (radius < 2) & (alive from dir < 1) -> (core, center) in dir; } state body { color 1 1 0 emits (alive, 1) transitions (alive from dir < 1) & (radius > 1) -> (body, body) in dir; }
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
17
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
18
state corner { color red emits (length, 8), (alive, 1) transitions (alive < 1) from dir -> (corner, segment) in dir; -> (corner); } state segment { color cyan emits (alive, 1) forwards (length - 1) transitions (length > 1.5) from dir & (alive < 0.5) from opposite (dir) -> (segment, segment) in opposite (dir); (length > 0.1) -> (corner); (length < 0.1) -> die; }
ID03 - 24 Sept 2003
Network Mesh
swarm.cs.virginia.edu
19
Composing Primitives
• Cells can follow multiple programs simultaneously (vector of independent states) • Cells can combine primitives through shared chemicals
– Chemicals secreted by one primitive can induce changes in other primitives
• Goals:
– Predict properties of composition based on properties of primitives – Diversity of primitive implementations provides protection from directed attacks
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
20
Mickey Mouse Program • 20 states • 50 transition rules • Starts from one cell, combines lines, spheres Real Mouse Program • 3B base pairs • 98% same as human DNA • Starts from one cell, combines complex proteins
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
21
Towards Real Systems
• Cells
– Sensor Devices, MEMS, Internet Nodes
• Division
– Processes – Find new hosts
• Communication
– Point-to-point emissions – Wireless multicast (can be multi-hop) diffusions
• Example: distributed file system running on simulated wireless nodes
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
22
Distributed Wireless File Service
File Distribution and Update
Publishe r
inhibit
publish
ID03 - 24 Sept 2003
swarm.cs.virginia.edu
23
Distributed Wireless File Service
File Distribution and Update
replicate
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
24
Robustness to Node Failures
80% of requests satisfied
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
25
Summary
• Trillions of creatures have died to evolve the extremely robust programs that survive today • Robustness and scalability require:
– Decentralization – Awareness of surroundings – Diversity
• Swarm Programming
– Develop high level behaviors from local interactions – Use communication through environment to coordinate locally – Produce complex behaviors by combining primitives defined over groups
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
26
Acknowledgements
Sponsor: National Science Foundation Contributors: Lance Davidson (UVa Biology) Selvin George Salvatore Guarnieri Steven Marchette Qi Wang Brian Zhang
http://swarm.cs.virginia.edu
ID03 - 24 Sept 2003 swarm.cs.virginia.edu
27
