Next Generation Network Science: An Overview

1
Next Generation Network Science An Overview
Michael Kearns and Ali Jadbabaie University of
Pennsylvania
2
Team Members
Dave Alderson Brian Stickler
Jean Carlson
Naval Postgraduate School
UC Santa Barbara
Michael Kearns (PI) Ali Jadbabaie Shawndra Hill
University of Pennsylvania
John Doyle Babak Hassibi
Fan Chung Graham
UC San Diego
Caltech
3

• Good news
• Spectacular progress
• Bad news
• Persistent errors and confusion
• Potentially insurmountable obstacles?

4
Challenges in the NS report

• Dynamics, spatial location, and information
  propagation in networks.
• Modeling and analysis of very large networks.
• Design and synthesis of networks.
• Increasing the level of rigor and mathematical
  structure.
• Abstracting common concepts across fields.
• Better experiments and measurements of network
  structure.
• Robustness and security of networks.

5
Challenges

• Goals
• Abstraction (common concepts across fields)
• Rigor ( math structure)
• Issues
• Dynamics (location, propagation)
• Robustness ( security)
• Levels of understanding
• 0. Verbal
• Data statistics (Experiments measurements)
• Modeling simulation
• Analysis
• Design synthesis

6
Theory and the Internet

• Goals
• Abstraction
• Rigor
• Issues
• Dynamics
• Robustness
• Levels
• 0. Verbal
• Data stats
• Modeling sim
• Analysis
• Design synth

Good news Spectacular progress

• Topics
• Traffic
• Topology
• Control and dynamics (CD)
• Layering/distributed
• Architecture

7
Huge and recent progress
8
Addressing challenges
Dynamics, Games, and New Models of Networks
Behavioral Network Science
Network Information theory
Network Architecture
Dynamics of Networks
project thrusts vs. Challenges
Novel Algorithms
9
Network of Investigators
Carlson
Doyle
Hassibi
Jadbabaie
10
Thrust 1 Novel Algorithms

• Local, Distributed graph Algorithms
  (Chung-Graham, Jadbabaie)
• Graph algorithms for partitioning
• Understanding role of Randomness , and Random
  graph models (Chung-Graham, Doyle, Carlson)
• Beyond degree distributions
• Matching and re-identification, data mining
  (Hill)
• Efficient scoring and identity matching

11
Thrust 2 Dynamics on Networks and Network Models

• Analysis and design of interconnected dynamical
  systems over networks, distributed optimization
  (Jadbabaie, Doyle, Hassibi)
• Global behaviors translated to local decisions
• Interplay of interconnection and dynamics
• Network formation games (Alderson, Kearns)
• New Models of Networks (Jadbabaie)
• From Graphs to Simplicial Complexes

12
Thrust 3Architecture

• Comparative Physiology of Network Architecture
  (Alderson, Doyle, Carlson)
• From Internet to biology
• Robustness, fragility and evolvability of complex
  networks
• Optimization , Layering, and games (Jadbabaie,
  Doyle, Alderson, Kearns)
• Layering as a tool for optimization decomposition

13
Thrust 4Network Information theory

• Entropic Vectors New tool for network
  information theory (Hassibi)
• Entropic vectors and convex optimization
• Fundamental limits in network information theory
  (Doyle, Hassibi)
• Connecting fundamental limits due to information,
  computation, and dynamics

14
Thrust 5 Behavioral Network Science

• Behavioral and Mathematical models for collective
  problem-solving (Kearns)
• Collective problem solving vs. distributed
  optimization (Kearns, Jadbabaie)

15
Thrust 6Testbeds and Demonstrations

• Hastily Formed Networks (Steckler, Alderson)
• Analysis of Field exercise data
• Measurement and statistics of field operations