Hybrid Unicast and Multicast Flow Control: A Linear Optimization Approach

Similar documents
A Link Transmission Model for Air Traffic Flow Prediction and Optimization

An Admission Control Algorithm in Cloud Computing Systems

Intra-Domain Traffic Engineering

ECE559VV Project Report

The strict priority scheduler

Priority Queuing with Finite Buffer Size and Randomized Push-out Mechanism

CS 331 DESIGN AND ANALYSIS OF ALGORITHMS DYNAMIC PROGRAMMING. Dr. Daisy Tang

Resource Allocation with a Budget Constraint for Computing Independent Tasks in the Cloud

EEL 6266 Power System Operation and Control. Chapter 3 Economic Dispatch Using Dynamic Programming

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009

Assortment Optimization under MNL

Rate Constrained Power Control in Space-Time Coded Fading Ad-Hoc Networks

Pricing Network Services by Jun Shu, Pravin Varaiya

Application of Nonbinary LDPC Codes for Communication over Fading Channels Using Higher Order Modulations

Queueing Networks II Network Performance

Lecture 2 Solution of Nonlinear Equations ( Root Finding Problems )

Embedded Systems. 4. Aperiodic and Periodic Tasks

Concepts for Wireless Ad Hoc

Computing Correlated Equilibria in Multi-Player Games

Tornado and Luby Transform Codes. Ashish Khisti Presentation October 22, 2003

Calculation of time complexity (3%)

Flexible Allocation of Capacity in Multi-Cell CDMA Networks

Rethinking MIMO for Wireless Networks: Linear Throughput Increases with Multiple Receive Antennas

Lecture 4: November 17, Part 1 Single Buffer Management

Tokyo Institute of Technology Periodic Sequencing Control over Multi Communication Channels with Packet Losses

On the Multicriteria Integer Network Flow Problem

: Numerical Analysis Topic 2: Solution of Nonlinear Equations Lectures 5-11:

Capacitor Placement In Distribution Systems Using Genetic Algorithms and Tabu Search

Amiri s Supply Chain Model. System Engineering b Department of Mathematics and Statistics c Odette School of Business

COMPLETE BUFFER SHARING IN ATM NETWORKS UNDER BURSTY ARRIVALS

Portfolios with Trading Constraints and Payout Restrictions

Single-Facility Scheduling over Long Time Horizons by Logic-based Benders Decomposition

Problem Set 9 Solutions

Natural Language Processing and Information Retrieval

Sketching Sampled Data Streams

Maximizing Overlap of Large Primary Sampling Units in Repeated Sampling: A comparison of Ernst s Method with Ohlsson s Method

An Integrated OR/CP Method for Planning and Scheduling

I + HH H N 0 M T H = UΣV H = [U 1 U 2 ] 0 0 E S. X if X 0 0 if X < 0 (X) + = = M T 1 + N 0. r p + 1

Resource Allocation for On-Demand Data Delivery to High-Speed Trains via Trackside Infostations

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Reduced-Dimensional MUSIC Algorithm for Joint Angle and Delay Estimation Based on L2 Norm Constraint in Multipath Environment

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017

Mathematical Economics MEMF e ME. Filomena Garcia. Topic 2 Calculus

Lecture 21: Numerical methods for pricing American type derivatives

Optimal information storage in noisy synapses under resource constraints

Global Optimization of Truss. Structure Design INFORMS J. N. Hooker. Tallys Yunes. Slide 1

An Interactive Optimisation Tool for Allocation Problems

Flexible Quantization

18.1 Introduction and Recap

Lecture 8: Time & Clocks. CDK: Sections TVS: Sections

Provable Security Signatures

Chapter Newton s Method

Numerical Methods Solution of Nonlinear Equations

Distributed Optimal TXOP Control for Throughput Requirements in IEEE e Wireless LAN

Outline. Communication. Bellman Ford Algorithm. Bellman Ford Example. Bellman Ford Shortest Path [1]

Retrieval Models: Language models

CISE301: Numerical Methods Topic 2: Solution of Nonlinear Equations

Timing-Driven Placement. Outline

Simultaneous Optimization of Berth Allocation, Quay Crane Assignment and Quay Crane Scheduling Problems in Container Terminals

TOPICS MULTIPLIERLESS FILTER DESIGN ELEMENTARY SCHOOL ALGORITHM MULTIPLICATION

C e n t r u m v o o r W i s k u n d e e n I n f o r m a t i c a

Topic 5: Non-Linear Regression

A FAST HEURISTIC FOR TASKS ASSIGNMENT IN MANYCORE SYSTEMS WITH VOLTAGE-FREQUENCY ISLANDS

Outline and Reading. Dynamic Programming. Dynamic Programming revealed. Computing Fibonacci. The General Dynamic Programming Technique

Planning and Scheduling to Minimize Makespan & Tardiness. John Hooker Carnegie Mellon University September 2006

The Minimum Universal Cost Flow in an Infeasible Flow Network

ECE 534: Elements of Information Theory. Solutions to Midterm Exam (Spring 2006)

Prof. Paolo Colantonio a.a

Vickrey Auction VCG Combinatorial Auctions. Mechanism Design. Algorithms and Data Structures. Winter 2016

Information Weighted Consensus

An Optimization Model for Routing in Low Earth Orbit Satellite Constellations

CHAPTER 4d. ROOTS OF EQUATIONS

NUMERICAL DIFFERENTIATION

Statistical tables are provided Two Hours UNIVERSITY OF MANCHESTER. Date: Wednesday 4 th June 2008 Time: 1400 to 1600

An Upper Bound on SINR Threshold for Call Admission Control in Multiple-Class CDMA Systems with Imperfect Power-Control

Exercises of Chapter 2

ExxonMobil. Juan Pablo Ruiz Ignacio E. Grossmann. Department of Chemical Engineering Center for Advanced Process Decision-making. Pittsburgh, PA 15213

Copyright 2017 by Taylor Enterprises, Inc., All Rights Reserved. Adjusted Control Limits for U Charts. Dr. Wayne A. Taylor

Moo-rings. Marina Mooring Optimization. Group 8 Route 64 Brian Siefering Amber Mazooji Kevin McKenney Paul Mingardi Vikram Sahney Kaz Maruyama

Minimisation of the Average Response Time in a Cluster of Servers

Equilibrium with Complete Markets. Instructor: Dmytro Hryshko

INTEGER NETWORK SYNTHESIS PROBLEM FOR HOP CONSTRAINED FLOWS

Pricing and Resource Allocation Game Theoretic Models

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan

Lecture 10 Support Vector Machines II

Goal Programming Approach to Solve Multi- Objective Intuitionistic Fuzzy Non- Linear Programming Models

Credit Card Pricing and Impact of Adverse Selection

Cognitive Access Algorithms For Multiple Access Channels

CS 3710: Visual Recognition Classification and Detection. Adriana Kovashka Department of Computer Science January 13, 2015

Modelling and Constraint Hardness Characterisation of the Unique-Path OSPF Weight Setting Problem

Coarse-Grain MTCMOS Sleep

Stat 543 Exam 2 Spring 2016

DETERMINATION OF TEMPERATURE DISTRIBUTION FOR ANNULAR FINS WITH TEMPERATURE DEPENDENT THERMAL CONDUCTIVITY BY HPM

Lecture 14: Bandits with Budget Constraints

Suggested solutions for the exam in SF2863 Systems Engineering. June 12,

En Route Traffic Optimization to Reduce Environmental Impact

IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 16, NO. 12, DECEMBER

Optimizing Resource Utilization in Wireless Multimedia Networks

AN IMPROVED METHOD OF HIERARCHIC ANALYSIS FOR CHOOSING OPTIMAL INFORMATION PROTECTION SYSTEM IN COMPUTER NETWORKS

MotionCast: On the Capacity and Delay Tradeoffs

Transcription:

Hybrd Uncast and Multcast Flow Control: A Lnear Optmzaton Approach Homayoun Youse zadeh Fatemeh Fazel Hamd Jaarhan Department o EECS Unversty o Calorna, Irvne

Flow Control Motvaton Content delvery over mutlcast networs s challengng Real-Tme Delay Constrants Recever Heterogenety Intra-Sesson Farness Qualty o Servce Inter-Sesson (Flow) Farness Networ Resource Sharng Flow arness or mxed UDP and TCP sessons s complcated. Unle TCP, UDP lacs bult-n low arness support. A generc soluton s requred.

Bacground Wor Farness Issues Ammar, Georga Tech Rubensten, Columba Kurose, UMass Feedbac Aggregaton Imploson Garca, UCSC DeLuca, USC Flow Control Srant, UIUC Kelly, Cambrdge Low, Caltech Jae, MIT Towsley, UMass Congeston Control Jacobson, UCB Floyd, UCB Vcsano, UC London

Flow Farness Utlty The utlty s dened as x U( x) = mn(, ) = where s the requested bandwdth o low wth... 2 x,, x x > Max-Mn Farness Denton: A bandwdth allocaton scheme among a number o competng lows s max-mn ar no low can be allocated a hgher bandwdth wthout reducng the allocaton o another low wth an equal or a lower rate.

Centralzed Flow Control: Global Topology Problem Formulaton x x, x, = max W ( ) Subject To : w x C j {...,, l} Further, = { W,..., W } { C,..., C l }... x {...,, } x x {...,, } } + 2 weghtng unctons ln capactes w j LP can be solved wth 2 complexty O(( l+ 2 )) j j, lowutlzesln j = 0, otherwse Theorem: The soluton s max-mn ar by W >... > W Multcast reward by settng W W η = = η where { η,..., η } are the number o low end nodes. j L w j

Decentralzed Flow Control: Local Ln Problem Formulaton x max x, x, = Subject To : x C = 2 C ln capacty... 2 Theorem: The LP can be solved wth lnear O( ) complexty utlzng the water-llng approach x x x Water-Fllng Soluton Case : C, then Case 2: C j, then nteger j = x =, < j =, h h x = C j= j, h+ h h C 0 h h = 0 h satses h+ 0 h = 0 j

Quas-Centralzed Flow Control: Zone Comparson o FC Schemes Global soluton may be overhead prohbtve Local Soluton s max-mn ar but sub-optmal or the overall topology Soluton Hybrd Flow Control Tradeo between accuracy and complexty Exchange state normaton wthn the local zones Mnmum ar share o a low spannng over multple zones s consdered Descrpton Dene local zones wth assocated Desgnated Nodes (DNs) Zone Dscovery Process DN sends broadcast plot pacets wth a gven TTL Eectve or sparse topologes wth ew populated areas Hybrd complexty o O l 2 (( z + 2 z)) wth l lns and z n the largest zone z lows

ECN Marng Utlze Bnary Marng Gven an ordered lst o per ln ar shares 2 n l l... l Identy bottlenec ln by conveyng λ λ Φ ++ = + ++ 2 n N n N N N N N = [( λ) ( λ ) ] ( λ)[ ( ) ( ) ] λ λ λ to end nodes, where l N Mar wth probablty exp( ( λ ) ) Measure the recevng rate o N unmared pacets Ξ = exp( Φ ) Estmate mnmum ar share as Requres numercal stablzaton λ = ( ln Ξ ) N

ECN-Based Implementaton: Uncast Intermedate Node Calculate mnmum ar share o the ln rom one o the decentralzed or quascentralzed FC algorthms Determne the values o C λ = /l = l Calculate lmax,, [ ] 2 ε b, and = l mn, l ε max, [ ] 2 ε l max, mn, ϕ C b λ = log ( ) End Node Calculate the rate o recevng unmared pacets or the prevous tme nterval as N Ξ = exp( Φ ) Approxmate mnmum ar share o the path rom the source as ( C / ( b ) ϑ ) where / ϑ [ ln )] N = Ξ Mar a pacet wth probablty N exp( ( ϕ ) ) or some large N

Multcastng Implcatons Consder one-to-many multcast trees Expand a multcast sesson as a set o vrtual uncast sessons wth the multcast source and an ndvdual recever. Address eedbac mploson n the dscovery process Feedbac aggregaton or detectng ndvdual ar shares n a group o nodes. Tmer-based eedbac supresson or detectng the overall mnmum ar share o a group o recevers.

Layered Meda Flow Control Treat each layer as an ndependent low. Snce, W... W and..., low-bandwdth hgh-prorty lows are accommodated rst under both global and local algorthms. Under partal ulllment due to bandwdth lmtaton, the bandwdth assgned to low-prorty groups s gven to hghprorty groups.

Topology (): Flow Control

Topology (2): Layered Meda

Than You http://newport.eecs.uc.edu/~hyouse