Recent developments in the study of transport in random networks Shai Carmi Bar-Ilan University Havlin group Minerva meeting Eilat, March 2009.

Slides:



Advertisements
Similar presentations
Complex Network Theory
Advertisements

Complex Networks Advanced Computer Networks: Part1.
Capacity of wireless ad-hoc networks By Kumar Manvendra October 31,2002.
Albert-László Barabási
Queuing Network Models for Delay Analysis of Multihop Wireless Ad Hoc Networks Nabhendra Bisnik and Alhussein Abouzeid Rensselaer Polytechnic Institute.
Modeling Malware Spreading Dynamics Michele Garetto (Politecnico di Torino – Italy) Weibo Gong (University of Massachusetts – Amherst – MA) Don Towsley.
It’s a Small World by Jamie Luo. Introduction Small World Networks and their place in Network Theory An application of a 1D small world network to model.
School of Information University of Michigan Network resilience Lecture 20.
VL Netzwerke, WS 2007/08 Edda Klipp 1 Max Planck Institute Molecular Genetics Humboldt University Berlin Theoretical Biophysics Networks in Metabolism.
Daniel ben -AvrahamClarkson University Boston Universtiy Reuven Cohen Tomer Kalisky Alex Rozenfeld Bar-Ilan University Eugene Stanley Lidia Braunstein.
Generated Waypoint Efficiency: The efficiency considered here is defined as follows: As can be seen from the graph, for the obstruction radius values (200,
Identity and search in social networks Presented by Pooja Deodhar Duncan J. Watts, Peter Sheridan Dodds and M. E. J. Newman.
CS 599: Social Media Analysis University of Southern California1 The Basics of Network Analysis Kristina Lerman University of Southern California.
Farnoush Banaei-Kashani and Cyrus Shahabi Criticality-based Analysis and Design of Unstructured P2P Networks as “ Complex Systems ” Mohammad Al-Rifai.
Cascading failures in interdependent networks and financial systems -- Departmental Seminar Xuqing Huang Advisor: Prof. H. Eugene Stanley Collaborators:
Weighted networks: analysis, modeling A. Barrat, LPT, Université Paris-Sud, France M. Barthélemy (CEA, France) R. Pastor-Satorras (Barcelona, Spain) A.
Hierarchy in networks Peter Náther, Mária Markošová, Boris Rudolf Vyjde : Physica A, dec
1 Evolution of Networks Notes from Lectures of J.Mendes CNR, Pisa, Italy, December 2007 Eva Jaho Advanced Networking Research Group National and Kapodistrian.
Emergence of Scaling in Random Networks Barabasi & Albert Science, 1999 Routing map of the internet
MEDUSA – New Model of Internet Topology Using k-shell Decomposition Shai Carmi Shlomo Havlin Bloomington 05/24/2005.
Zhenhua Wu Advisor: H. E. StanleyBoston University Co-advisor: Lidia A. BraunsteinUniversidad Nacional de Mar del Plata Collaborators: Shlomo HavlinBar-Ilan.
1 University of Freiburg Computer Networks and Telematics Prof. Christian Schindelhauer Mobile Ad Hoc Networks Theory of Data Flow and Random Placement.
Physical Mechanism Underlying Opinion Spreading
1 Algorithms for Large Data Sets Ziv Bar-Yossef Lecture 7 May 14, 2006
Large-scale organization of metabolic networks Jeong et al. CS 466 Saurabh Sinha.
The Erdös-Rényi models
Optimization Based Modeling of Social Network Yong-Yeol Ahn, Hawoong Jeong.
Epidemic spreading in complex networks: from populations to the Internet Maziar Nekovee, BT Research Y. Moreno, A. Paceco (U. Zaragoza) A. Vespignani (LPT-
(Social) Networks Analysis III Prof. Dr. Daning Hu Department of Informatics University of Zurich Oct 16th, 2012.
Developing Analytical Framework to Measure Robustness of Peer-to-Peer Networks Niloy Ganguly.
IEEE P2P, Aachen, Germany, September Ad-hoc Limited Scale-Free Models for Unstructured Peer-to-Peer Networks Hasan Guclu
Claudio Castellano CNR-INFM Statistical Mechanics and Complexity and
DATA MINING LECTURE 13 Absorbing Random walks Coverage.
MEDUSA – New Model of Internet Topology Using k-shell Decomposition Shai Carmi Shlomo Havlin Bloomington 05/24/2005.
Weighted networks: analysis, modeling A. Barrat, LPT, Université Paris-Sud, France M. Barthélemy (CEA, France) R. Pastor-Satorras (Barcelona, Spain) A.
Network Flow How to solve maximal flow and minimal cut problems.
Random-Graph Theory The Erdos-Renyi model. G={P,E}, PNP 1,P 2,...,P N E In mathematical terms a network is represented by a graph. A graph is a pair of.
Self-Similarity of Complex Networks Maksim Kitsak Advisor: H. Eugene Stanley Collaborators: Shlomo Havlin Gerald Paul Zhenhua Wu Yiping Chen Guanliang.
Social Network Analysis Prof. Dr. Daning Hu Department of Informatics University of Zurich Mar 5th, 2013.
Physics of Flow in Random Media Publications/Collaborators: 1) “Postbreakthrough behavior in flow through porous media” E. López, S. V. Buldyrev, N. V.
Efficient Labeling Scheme for Scale-Free Networks The scheme in detailsPerformance of the scheme First we fix the number of hubs (to O(log(N))) and show.
Analyzing the Vulnerability of Superpeer Networks Against Attack Niloy Ganguly Department of Computer Science & Engineering Indian Institute of Technology,
Complex Network Theory – An Introduction Niloy Ganguly.
Robustness and Structure of Complex Networks Shuai Shao Boston University, Physics Department A dvisor: Prof. Gene Stanley Co-advisor: Prof. Shlomo Havlin.
Workshop on Optimization in Complex Networks, CNLS, LANL (19-22 June 2006) Application of replica method to scale-free networks: Spectral density and spin-glass.
Class 9: Barabasi-Albert Model-Part I
Lecture 10: Network models CS 765: Complex Networks Slides are modified from Networks: Theory and Application by Lada Adamic.
KPS 2007 (April 19, 2007) On spectral density of scale-free networks Doochul Kim (Department of Physics and Astronomy, Seoul National University) Collaborators:
Application of statistical physics to random graph models of networks Sameet Sreenivasan Advisor: H. Eugene Stanley.
Percolation and diffusion in network models Shai Carmi, Department of Physics, Bar-Ilan University Networks Percolation Diffusion Background picture: The.
Brief Announcement : Measuring Robustness of Superpeer Topologies Niloy Ganguly Department of Computer Science & Engineering Indian Institute of Technology,
March 3, 2009 Network Analysis Valerie Cardenas Nicolson Assistant Adjunct Professor Department of Radiology and Biomedical Imaging.
Condensation in/of Networks Jae Dong Noh NSPCS08, 1-4 July, 2008, KIAS.
Fractional Feynman-Kac Equation for non-Brownian Functionals IntroductionResults Applications See also: L. Turgeman, S. Carmi, and E. Barkai, Phys. Rev.
Transport in weighted networks: optimal path and superhighways Collaborators: Z. Wu, Y. Chen, E. Lopez, S. Carmi, L.A. Braunstein, S. Buldyrev, H. E. Stanley.
Load Balanced Link Reversal Routing in Mobile Wireless Ad Hoc Networks Nabhendra Bisnik, Alhussein Abouzeid ECSE Department RPI Costas Busch CSCI Department.
Urban Traffic Simulated From A Dual Perspective Hu Mao-Bin University of Science and Technology of China Hefei, P.R. China
Path Planning Based on Ant Colony Algorithm and Distributed Local Navigation for Multi-Robot Systems International Conference on Mechatronics and Automation.
Percolation Percolation is a purely geometric problem which exhibits a phase transition consider a 2 dimensional lattice where the sites are occupied with.
Netlogo demo. Complexity and Networks Melanie Mitchell Portland State University and Santa Fe Institute.
Response network emerging from simple perturbation Seung-Woo Son Complex System and Statistical Physics Lab., Dept. Physics, KAIST, Daejeon , Korea.
Mean Field Methods for Computer and Communication Systems Jean-Yves Le Boudec EPFL Network Science Workshop Hong Kong July
Structures of Networks
Hiroki Sayama NECSI Summer School 2008 Week 2: Complex Systems Modeling and Networks Network Models Hiroki Sayama
Random walks on complex networks
Hidden Markov Models Part 2: Algorithms
Shortest path and small world effect
Department of Computer Science University of York
Peer-to-Peer and Social Networks
Log-periodic oscillations due to discrete effects in complex networks
Presentation transcript:

Recent developments in the study of transport in random networks Shai Carmi Bar-Ilan University Havlin group Minerva meeting Eilat, March 2009

Networks Why do we care about networks? Networks appear everywhere: Communication (Internet, p2p,…) Transportation (roads, airlines,…). Social sciences (social networks, business relations,…) Life sciences (gene regulation, food webs,…)

Transport in networks Networks are commonly used as a platform for transport of: * Information (communication and social networks) * Passengers and commodities (transportation networks) * Current (electric circuits) * Diseases (social networks) Quantities of interest: * Time to reach target* Maximal capacity * Number of links crossed* Congestion and avalanches * Load at each node* Diffusion coefficients

Network models 1. Most naïve model: a regular lattice. * Only good for purely spatial, local, interactions. 2. Erdos-Renyi (ER) network model: fully random. * Fixed number of nodes N, each link exists with probability p. * Narrow degree distribution: where k is node degree. 3. Scale-free (SF) networks: emergence of hubs. * Broad degree distribution: * Nodes with extremely high degree exist (hubs). * Other ingredients possible, e.g., growth, correlations. * Found to describe most real-world systems.

Transport models: outline 1.Random walk with priorities. 2.Random walk with trapping. 3.Maximum flow.

Transport models 1.Random walk with priorities. 2.Random walk with trapping. 3.Maximum flow.

Motivation Some communication networks use random walk to search for other computers or spread information. Some data packets have higher priority than others. How does priority policy affect the diffusion in the network?

Two species of particles, A and B with densities ρ A and ρ B. A is high priority, B is low priority. Symmetric random walk (nearest neighbors). Protocols B can move only after all the A’s in its site have already moved. If motion is impossible, choose again. Model definition AB Site protocol: A site is randomly chosen and sends a particle. Particle protocol: A particle is randomly chosen and jumps out.

Solution in lattices Write a Markov chain for the number of particles in a site. Solve for the stationary probabilities. Derive analytically the fraction of empty sites in both protocols. Diffusion is normal: =Dt. Apply the site protocol selection rule and find D for each species. In the particle protocol define r as the fraction of free B's to total B’s. independent of ρ B and approaches for large densities. Derive diffusion coefficients similarly.

Solution in networks In the particle protocol: or, the probability of a site to be empty decreases exponentially with its degree. In scale-free networks, A’s move freely, and tend to aggregate at the hubs. Therefore, B’s at the hubs have very low probability to escape. Since the B’s themselves are attracted to the hubs, they eventually become trapped and their motion is arrested. Time for a B to leave a site of degree k. Waiting time distribution → sub-diffusion. Lattice, ER SF SF,ER Real Internet Average waiting time for B particles. Distribution of waiting times for B particles.

Transport models 1.Random walk with priorities. 2.Random walk with trapping. 3.Maximum flow.

Motivation Consider again a random walk process in a network. In a communication or a social network, a message can disappear; for example, due to failure. How long will the message survive before being trapped?

Particles initially evenly distributed over the network. Symmetric random walk (nearest neighbors). m of the nodes are absorbing. Whenever the particle reaches the trap it is absorbed. What is the survival probability ρ(t)? Model definition

A simple theory Denote the total number of links entering the traps by k m. The total number of links is N. Thus, the probability per unit time of a particle to enter the trap is approximately proportional to k m /N., and the problem is reduced to evaluating k m for different topologies. In ER networks: - Approximation is good when. - Explicit dependence on both m and N. For dense enough SF networks: -k min is the minimum degree (one trap).

Results The average time before trapping T usually scales as N. In SF networks when one of the hubs is a trap -Only for infinite γ SF and ER networks are equivalent. SF networks become less vulnerable as links are added. For ER networks A=1-1/. Conclusion: A simple mean-field approach is usually useful to solve trapping problem in networks, and leads to interesting observations. Theory- lines Simulation- symbols ER

Transport models 1.Random walk with priorities. 2.Random walk with trapping. 3.Maximum flow.

Motivation Users in communication networks (e.g., peer-to-peer) wish to exchange files by sending them through the network links. How many users can exchange files without interfering with each other? What is the maximum capacity of the network for a given number of users?

Assume the network contains n sources and n sinks. Consider three types of transport: * Maximum flow (= #of parallel paths) * Electric current * Multi-commodity flow Non directed, non weighted (unit capacities/ resistances). Model definition n Sinks T2T2 T1T1 n Sources S2S2 S1S1 Rest of network Regular flow Multi-commodity flow

Theory for small n For a single source/sink pair with degrees k 1 and k 2, F≈min(k 1,k 2 ). For small n, replace k 1 by the total number of links leaving the sources, and similarly for k 2. The distribution of flows is: * For ER networks: * For SF networks: Flow per user increases with n up to the optimal number of users above which the approximation is invalid. Small n theory Simulations

Theory for large n A different approach is needed: Find the total flow by conditioning on the number of paths of a given length. F = F 1 + F 2 + F 3 + … I ≈ F 1 /1 + F 2 /2 + F 3 /3 + … For direct linkage, =n 2 p. Implicit sum formulas for,. Sinks Sources F2F2 F3F3 F1F1 Theory- lines Simulation- symbols ER networks

Multi-commodity flow Flow from a source is directed towards specific sink. Thus the contribution of the different source/sink pairs can be separated. Result for ER network: where k n is the effective degree of the network when n pairs communicate. The network will saturate at the percolation threshold, when and thus. In SF networks the absence of percolation threshold leads to increased capacity. Small n approximation

Summary Transport in networks is important and interesting. We introduced models inspired by problems in real networks. We used probability theory and computer simulations to obtain analytical an numerical solutions. Deep relations between network structure and dynamics were uncovered. For example: * Halting of low priority particles in highly connected nodes. * Effect of failure in hubs on particle survival probability. * Optimal number of users in flow network. * Influence of inter-node distance on electrical current. * Interplay between percolation theory and maximal network flow.

Collaborators My advisor: Prof. Shlomo Havlin, Bar Ilan Univ., Israel. Other collaborators: Prof. Daniel ben-Avraham (Clarkson Univ., NY, USA) Prof. Panos Argyrakis (Aristotle Univ., Thessaloniki, Greece) Prof. H. Eugene Stanley (Boston Univ., MA, USA). ShlomoDaniPanosGene

Thank you for your attention! See also: M. Maragakis, S. Carmi, D. ben-Avraham, S. Havlin, and P. Argyrakis. "Priority diffusion model in lattices and complex networks". Phys. Rev. E (RC) 77, (2008). S. Carmi, Z. Wu, S. Havlin, and H. E. Stanley. "Transport in networks with multiple sources and sinks". Europhys. Lett. 84, (2008). A. Kittas, S. Carmi, S. Havlin, and P. Argyrakis. "Trapping in complex networks“. Europhys. Lett. 84, (2008).