Design Issues of Reserved Delivery Subnetworks Ruibiao Qiu Department of Computer Science and Engineering Washington University in St. Louis April 28,

Slides:



Advertisements
Similar presentations
Ch. 12 Routing in Switched Networks
Advertisements

Lecture 7. Network Flows We consider a network with directed edges. Every edge has a capacity. If there is an edge from i to j, there is an edge from.
Impact of Interference on Multi-hop Wireless Network Performance Kamal Jain, Jitu Padhye, Venkat Padmanabhan and Lili Qiu Microsoft Research Redmond.
~1~ Infocom’04 Mar. 10th On Finding Disjoint Paths in Single and Dual Link Cost Networks Chunming Qiao* LANDER, CSE Department SUNY at Buffalo *Collaborators:
1 Advancing Supercomputer Performance Through Interconnection Topology Synthesis Yi Zhu, Michael Taylor, Scott B. Baden and Chung-Kuan Cheng Department.
1 Efficient and Robust Streaming Provisioning in VPNs Z. Morley Mao David Johnson Oliver Spatscheck Kobus van der Merwe Jia Wang.
1 EL736 Communications Networks II: Design and Algorithms Class8: Networks with Shortest-Path Routing Yong Liu 10/31/2007.
1 Estimating Shared Congestion Among Internet Paths Weidong Cui, Sridhar Machiraju Randy H. Katz, Ion Stoica Electrical Engineering and Computer Science.
Distributed Algorithms for Secure Multipath Routing
Mohamed Hefeeda 1 School of Computing Science Simon Fraser University, Canada Multimedia Streaming in Dynamic Peer-to-Peer Systems and Mobile Wireless.
2001 Winter CS215 Course Project Simulation Comparison of Routing Algorithms for Multicast with Bandwidth Reservation Zhihong Duan
Placement of Integration Points in Multi-hop Community Networks Ranveer Chandra (Cornell University) Lili Qiu, Kamal Jain and Mohammad Mahdian (Microsoft.
Traffic Engineering Jennifer Rexford Advanced Computer Networks Tuesdays/Thursdays 1:30pm-2:50pm.
More routing protocols Alec Woo June 18 th, 2002.
Peer-to-Peer Based Multimedia Distribution Service Zhe Xiang, Qian Zhang, Wenwu Zhu, Zhensheng Zhang IEEE Transactions on Multimedia, Vol. 6, No. 2, April.
Network Optimization Models: Maximum Flow Problems In this handout: The problem statement Solving by linear programming Augmenting path algorithm.
Study of Distance Vector Routing Protocols for Mobile Ad Hoc Networks Yi Lu, Weichao Wang, Bharat Bhargava CERIAS and Department of Computer Sciences Purdue.
Promoting the Use of End-to-End Congestion Control & Random Early Detection of Network Congestion.
S. Suri, M, Waldvogel, P. Warkhede CS University of Washington Profile-Based Routing: A New Framework for MPLS Traffic Engineering.
Multipath Routing Algorithms for Congestion Minimization Ron Banner and Ariel Orda Department of Electrical Engineering Technion- Israel Institute of Technology.
On Self Adaptive Routing in Dynamic Environments -- A probabilistic routing scheme Haiyong Xie, Lili Qiu, Yang Richard Yang and Yin Yale, MR and.
Connected Dominating Sets in Wireless Networks My T. Thai Dept of Comp & Info Sci & Engineering University of Florida June 20, 2006.
1 Algorithms for Bandwidth Efficient Multicast Routing in Multi-channel Multi-radio Wireless Mesh Networks Hoang Lan Nguyen and Uyen Trang Nguyen Presenter:
UCSC 1 Aman ShaikhICNP 2003 An Efficient Algorithm for OSPF Subnet Aggregation ICNP 2003 Aman Shaikh Dongmei Wang, Guangzhi Li, Jennifer Yates, Charles.
Network Coding vs. Erasure Coding: Reliable Multicast in MANETs Atsushi Fujimura*, Soon Y. Oh, and Mario Gerla *NEC Corporation University of California,
The Effects of Systemic Packets Loss on Aggregate TCP Flows Thomas J. Hacker May 8, 2002 Internet 2 Member Meeting.
Package Transportation Scheduling Albert Lee Robert Z. Lee.
MGR: Multi-Level Global Router Yue Xu and Chris Chu Department of Electrical and Computer Engineering Iowa State University ICCAD
Distributed Quality-of-Service Routing of Best Constrained Shortest Paths. Abdelhamid MELLOUK, Said HOCEINI, Farid BAGUENINE, Mustapha CHEURFA Computers.
Bargaining Towards Maximized Resource Utilization in Video Streaming Datacenters Yuan Feng 1, Baochun Li 1, and Bo Li 2 1 Department of Electrical and.
On the Construction of Data Aggregation Tree with Minimum Energy Cost in Wireless Sensor Networks: NP-Completeness and Approximation Algorithms National.
QoS-Aware In-Network Processing for Mission-Critical Wireless Cyber-Physical Systems Qiao Xiang Advisor: Hongwei Zhang Department of Computer Science Wayne.
The Minimal Communication Cost of Gathering Correlated Data over Sensor Networks EL 736 Final Project Bo Zhang.
07/21/2005 Senmetrics1 Xin Liu Computer Science Department University of California, Davis Joint work with P. Mohapatra On the Deployment of Wireless Sensor.
Network Aware Resource Allocation in Distributed Clouds.
1 Min-Cost Live Webcast under Joint Pricing of Data, Congestion and Virtualized Servers Rui Zhu 1, Di Niu1, Baochun Li 2 1 Department of Electrical and.
IEEE Globecom 2010 Tan Le Yong Liu Department of Electrical and Computer Engineering Polytechnic Institute of NYU Opportunistic Overlay Multicast in Wireless.
When In-Network Processing Meets Time: Complexity and Effects of Joint Optimization in Wireless Sensor Networks Department of Computer Science, Wayne State.
RANI NALAMARU DEPARTMENT OF COMPUTER SCIENCE BALL STATE UNIVERSITY RANI NALAMARU DEPARTMENT OF COMPUTER SCIENCE BALL STATE UNIVERSITY Efficient Transmission.
Energy-Aware Scheduling with Quality of Surveillance Guarantee in Wireless Sensor Networks Jaehoon Jeong, Sarah Sharafkandi and David H.C. Du Dept. of.
Boundary Recognition in Sensor Networks by Topology Methods Yue Wang, Jie Gao Dept. of Computer Science Stony Brook University Stony Brook, NY Joseph S.B.
Performance evaluation of video transcoding and caching solutions in mobile networks Jim Roberts (IRT-SystemX) joint work with Salah Eddine Elayoubi (Orange.
De-Nian Young Ming-Syan Chen IEEE Transactions on Mobile Computing Slide content thanks in part to Yu-Hsun Chen, University of Taiwan.
1 Network Coding and its Applications in Communication Networks Alex Sprintson Computer Engineering Group Department of Electrical and Computer Engineering.
1 Exploring Custom Instruction Synthesis for Application-Specific Instruction Set Processors with Multiple Design Objectives Lin, Hai Fei, Yunsi ACM/IEEE.
Engineering Jon Turner Computer Science & Engineering Washington University Coarse-Grained Scheduling for Multistage Interconnects.
InterConnection Network Topologies to Minimize graph diameter: Low Diameter Regular graphs and Physical Wire Length Constrained networks Nilesh Choudhury.
On the Optimal Scheduling for Media Streaming in Data-driven Overlay Networks Meng ZHANG with Yongqiang XIONG, Qian ZHANG, Shiqiang YANG Globecom 2006.
EE 685 presentation Optimization Flow Control, I: Basic Algorithm and Convergence By Steven Low and David Lapsley.
Multiuser Receiver Aware Multicast in CDMA-based Multihop Wireless Ad-hoc Networks Parmesh Ramanathan Department of ECE University of Wisconsin-Madison.
1 - CS7701 – Fall 2004 Review of: Detecting Network Intrusions via Sampling: A Game Theoretic Approach Paper by: – Murali Kodialam (Bell Labs) – T.V. Lakshman.
Flow and Congestion Control for Reliable Multicast Communication In Wide-Area Networks Supratik Bhattacharyya Department of Computer Science University.
CS223 Advanced Data Structures and Algorithms 1 Maximum Flow Neil Tang 3/30/2010.
1 Slides by Yong Liu 1, Deep Medhi 2, and Michał Pióro 3 1 Polytechnic University, New York, USA 2 University of Missouri-Kansas City, USA 3 Warsaw University.
November 4, 2003Applied Research Laboratory, Washington University in St. Louis APOC 2003 Wuhan, China Cost Efficient Routing in Ad Hoc Mobile Wireless.
Efficient Resource Allocation for Wireless Multicast De-Nian Yang, Member, IEEE Ming-Syan Chen, Fellow, IEEE IEEE Transactions on Mobile Computing, April.
Load Balanced Link Reversal Routing in Mobile Wireless Ad Hoc Networks Nabhendra Bisnik, Alhussein Abouzeid ECSE Department RPI Costas Busch CSCI Department.
1 Traffic Engineering By Kavitha Ganapa. 2 Introduction Traffic engineering is concerned with the issue of performance evaluation and optimization of.
Internet Traffic Engineering Motivation: –The Fish problem, congested links. –Two properties of IP routing Destination based Local optimization TE: optimizing.
Courtesy Piggybacking: Supporting Differentiated Services in Multihop Mobile Ad Hoc Networks Wei LiuXiang Chen Yuguang Fang WING Dept. of ECE University.
1 Chapter 5 Branch-and-bound Framework and Its Applications.
1 Transport Bandwidth Allocation 3/29/2012. Admin. r Exam 1 m Max: 65 m Avg: 52 r Any questions on programming assignment 2 2.
::Network Optimization:: Minimum Spanning Trees and Clustering Taufik Djatna, Dr.Eng. 1.
-1/16- Maximum Battery Life Routing to Support Ubiquitous Mobile Computing in Wireless Ad Hoc Networks C.-K. Toh, Georgia Institute of Technology IEEE.
Minimum Power Configuration in Wireless Sensor Networks Guoliang Xing*, Chenyang Lu*, Ying Zhang**, Qingfeng Huang**, and Robert Pless* *Washington University.
Confidential & Proprietary – All Rights Reserved Internal Distribution, October Quality of Service in Multimedia Distribution G. Calinescu (Illinois.
Impact of Interference on Multi-hop Wireless Network Performance
Presented by Tae-Seok Kim
Introduction Basic formulations Applications
Hemant Kr Rath1, Anirudha Sahoo2, Abhay Karandikar1
Presentation transcript:

Design Issues of Reserved Delivery Subnetworks Ruibiao Qiu Department of Computer Science and Engineering Washington University in St. Louis April 28, 2005

2 - Ruibiao Qiu - 4/28/2005 Motivations Lack of bandwidth reservation in today’s Internet »Dominant best-effort traffic »Why per-flow reservation not deployed? l Cost of equipment, management and operation l Support in applications and end hosts

3 - Ruibiao Qiu - 4/28/2005 Motivation (cont.) Aggregate bandwidth reservation: an alternative solution »Exclusive bandwidth reservation for aggregate of users »Manageable deployment for network providers »No change in applications and end hosts Numerous real world applications »Content delivery networks »Virtual private networks »Grid computing

4 - Ruibiao Qiu - 4/28/2005 Reserved Delivery Subnetwork (RDS) A mechanism to provide better service for large aggregates of users Larger aggregate is more efficient 100 flows 10,000 flows bursty flows (peak/avg=25) aggregate flow Per-flow reservation / average Overload probability

5 - Ruibiao Qiu - 4/28/2005 An RDS Example 120Mb/s 70Mb/s 15Mb/s10Mb/s

6 - Ruibiao Qiu - 4/28/2005 Example RDS Source node Sink nodeOther node

7 - Ruibiao Qiu - 4/28/2005 Outline Introduction Configuration of RDSs with single server Source traffic regulation in an RDS Summary

8 - Ruibiao Qiu - 4/28/2005 Single-server RDS Configuration Bandwidth reservation for an RDS »Satisfy all user demands »Use bandwidth efficiently Formulated as a graph problem s 5,35,3 4,34,3 5,15,1 9,19,1 9,49,4 3,23,2 4,44,4 7,27,2 7,27,2 5,25,2 8,58,5 8,38,3 1 2,42,4 2 5 capacity,length 2,22,2 sink demand 4 reservation

9 - Ruibiao Qiu - 4/28/2005 Problem Formulation Transformation to a network flow problem »Flow: average aggregate traffic on a link »Link cost = reservation x length »An RDS corresponds a minimum cost maximum flow s 5,35,3 4,34,3 5,15,1 9,19,1 9,49,4 3,23,2 4,44,4 7,27,2 7,27,2 5,25,2 8,58,5 8,38,3 t 1,01,0 2,02,0 5,05,0 1 2,42,4 2 5 capacity,length 2,22,2 sink demand 4 reservation ,42,4 flow ,21,2 7,87,8 1,21,2 5,75,7 5,75,7 5,75,7 1,21,2 flow=8,total cost=101 flow,reservation flow=8,total cost=75 flow ,21,2 5,85,8 8,98,98,98,9 1,21,2 flow,reservation

10 - Ruibiao Qiu - 4/28/2005 Bandwidth economy of aggregation »Larger flow aggregate, smaller fraction of traffic variation »Individual flows as i.i.d. random variables {X 1, X 2, …, X n }, and aggregate flow as X =  X i l  =  i,  = (  i 2 ) 1/2 »A concave function l Reservation grows more slowly than flow size Concave link cost function C(  f ) = l  (  f + k  f ) Link Cost Function link length reservation C(  f ) ff 0

11 - Ruibiao Qiu - 4/28/2005 Min-cost Max-flow Problem Find min. cost flow among all max. flows Efficient algorithms exist for linear cost networks For concave cost networks »NP-hard problems [Guisewite-Pardalos 1990] »Search-based exact algorithms impractical »Efficient approximation algorithms needed

12 - Ruibiao Qiu - 4/28/2005 Least Cost Augmentation (LCA) s t Optimal solutions in linear costs networks Saturate path with least incremental cost unit cost

13 - Ruibiao Qiu - 4/28/2005 Challenge: Concave Link Cost Effects Incremental cost »Linear cost links: linear to flow increment »Concave cost links: depends on current flow & flow increment Same initial flow, different flow increments  different augmentation paths incremental cost Flow increment = 1 Flow increment =

14 - Ruibiao Qiu - 4/28/2005 Largest Demand First (LDF) Approximate LCA in concave costs networks Largest sink demand as flow increment s t incremental cost incremental cost

15 - Ruibiao Qiu - 4/28/2005 Evaluation of LDF Simulations topologies »Torus network »National network (50 metro areas) »Random source and ≤50 sinks Variables »Number of sinks »Flow variations s

16 - Ruibiao Qiu - 4/28/2005 Maximum Sink Sharing Sort sinks by their distances to source Assume all sinks share a single (unrealistically) path to the source A loose bound s

17 - Ruibiao Qiu - 4/28/2005 Estimated Lower Bounds Equally partition nodes on a “disc” geographically Sort sinks by distance in each partition Assume all sinks share a single path to source s

18 - Ruibiao Qiu - 4/28/2005 Performance Evaluation Solutions evaluated »LDF l Largest demand first »EB(n) l Estimated lower bound with n “slices” »SPT l Shortest path tree as approximation »SPT(C) l Assuming no “incidental sharing” (star network) l Provide an upper bound Measure relative cost to lower bound (EB(1))

19 - Ruibiao Qiu - 4/28/2005 Simulation Results LDF within a constant factor of lower bound

20 - Ruibiao Qiu - 4/28/2005 Example RDS Source node Sink nodeOther node

21 - Ruibiao Qiu - 4/28/2005 A Local Search Approach Local search »Find the local optimal with efficient operations from a solution »An effective approximation method for combinatorial problems Using local search for local optimal solutions »Further improve solution quality »Measure the optimality of LDF solutions

22 - Ruibiao Qiu - 4/28/2005 Negative Cost Cycles Undirectional cycles Redirecting flow along the cycle  Negative total incremental cost After redirection, incremental cost = -3, lower cost solution flow incremental cost 1 redirected flow

23 - Ruibiao Qiu - 4/28/2005 Cycle Reduction An efficient operation for local search Must work in concave cost links v x w u non-flow edge 1 flow 3 5 w v x u s non-flow edge v x w u flow 2 6 1

24 - Ruibiao Qiu - 4/28/2005 Bicycles Negative bicycles in concave cost graphs Reduction leads to further cost improvements New discovery a b 1000 path distance redirected flow current flow edge cost = l (f+f 1/2 ) original cost: 8800 after redirection: 8700

25 - Ruibiao Qiu - 4/28/2005 Simulation Results Limited improvements by local searches Performance of LDF sufficient

26 - Ruibiao Qiu - 4/28/2005 Contributions Study precise aggregate bandwidth reservation in an RDS Formulate the network design problem as a minimum cost network flow problem Introduce more realistic concave cost functions Propose an efficient approximation solution for the NP-Hard problem Develop local search improvements with cycle and bicycle reduction

27 - Ruibiao Qiu - 4/28/2005 Outline Introduction Configuration of RDS with a single server Traffic regulation in an RDS Summary

28 - Ruibiao Qiu - 4/28/2005 End-to-end Performance Potentials Performance limitation in ordinary Internet RDS makes end-to-end performance improvements possible »Informed data transfer [Savage et al 99] »Knowledge about underlying network »Information about the data backlog at both ends Example »Solving unbalanced bandwidth utilization problem

29 - Ruibiao Qiu - 4/28/2005 Unbalanced Bandwidth Utilization Caused by overloaded sink »Overload some paths »Leave other paths under utilized Avoidable in an RDS Total reservation Actual usage Unused Source Sink Overloaded Under utilized

30 - Ruibiao Qiu - 4/28/2005 Source Traffic Regulation Source schedules data transfers to maximize bandwidth utilization Data transfer scheduling algorithm »Estimate sink draining time »Order sinks by increasing order of draining time »Always allow the fastest draining sink to send with maximum allowed rate

31 - Ruibiao Qiu - 4/28/2005 Per-connection Regulation Favor the less congested sinks sinks B i (1) R R1R1 C3C3 C1C1 C2C2 C3C3 C2C2 C1C1 B o (1) B i (5) B i (4) B i (3) B i (2) R(2) R(1) R1R1 R(3) R(4) R(5) B o (2) B o (3) B o (4) B o (5) r1r1 r2r2 r3r3 r4r4 r5r5 source Order: 1, 2, 3, 4, 5 Order: 2, 3, 4, 5, 1

32 - Ruibiao Qiu - 4/28/2005 Aggregated Regulation Per-connection traffic regulation overhead Aggregated information for feedback control RDS B i (1) B i (3) B i (2) R(1) R(2) R(3) sink source

33 - Ruibiao Qiu - 4/28/2005 Simulations Three sinks 100 flows/sink Avg. 1Mb/s per flow After 10s, one sink has additional 700 flows 400Mb/s 200Mb/s 300Mb/s 200Mb/s

34 - Ruibiao Qiu - 4/28/2005 Simulation Results Improve fairness, penalize overloaded sinks

35 - Ruibiao Qiu - 4/28/2005 Summary RDS: an effective alternative to per-flow reservation »Improved quality of service for aggregate of users »Easy to implement »Compatible with existing applications Research focus: RDS design issues »Configuration of single-server RDS »Configuration of multi-server RDS »Source traffic regulation inside RDS

36 - Ruibiao Qiu - 4/28/2005 Previous Research Projects ALX (adaptation layer translator)-based studio quality video conferencing system over broad band WAN (ICME02,GLOBECOM02) Studies of Motion JPEG2000 and its applications in video processing and multimedia communications (EI02,EI03) Quality-scalable Motion-JPEG2000 video streaming over active networks (EI03) Cost-based routing in ad hoc wireless networks Contributions to ACE code base ATM stream interfaces on Windows and Solaris

37 - Ruibiao Qiu - 4/28/2005 Questions?

38 - Ruibiao Qiu - 4/28/2005 Bicycle Reduction Redirect flow on two cycles r v x w y Non-tree vertices f(q,y) p q f(p,x) u r v x w y f(q,y) p q f(u,p)-f(p,x) f(p,x) f(v,w)+f(q,y)+f(p,x) u f(u,q)-f(q,y t v u x w to tree vertices In-tree path cost s non-tree edge cost y

39 - Ruibiao Qiu - 4/28/2005 An Extreme Bicycle Example No negative single cycle Improvement up to n s n n-1 n-2 m 1 link distance n-1 n n-2 m 1 1+ 

40 - Ruibiao Qiu - 4/28/2005 Simulation Results Limited improvements by local searches Performance of LDF sufficient

41 - Ruibiao Qiu - 4/28/2005 Simulation Results Improve fairness, penalize overloaded sinks