SlideShare una empresa de Scribd logo
1 de 37
Descargar para leer sin conexión
Summer 2008 Internship
Report

    Advisor: Prof. Angela Y. Zhang
      The Chinese University of Hong Kong, Hong Kong




          Student: Pratik Poddar
        Indian Institute of Technology Bombay, India




    Topic: Non­Convex Optimization 
         Problems in Networks
                               
Introduction


3 topics to be discussed:
 1) Polyblock Algorithm for Monotonic Optimization
 2) Network Utility Maximization
 3) Internet Congestion Control Problem




                          
Basics of Optimization
    ●
        Standard Optimization problem
    ●
        Linear Optimization problem
    ●
        Convex Optimization problem
    ●
        Monotonic Optimization problem




                                
Polyblock Algorithm
    ●
        We have had two major events in the history of 
        optimization theory.
    ●
         The first was linear programming and simplex 
        method in late 1940s­ early 1950s.
    ●
        The second was convex optimization and interior 
        point method in late 1980s­ early 1990s. 



                                 
Polyblock Algorithm
    ●
        Convex optimization problems are known to be 
        solved, very reliably and efficiently.
    ●
        "..in fact, the great watershed in optimization isn't 
        between linearity and nonlinearity, but convexity 
        and nonconvexity" ­ R. Tyrrell Rockafellar, in  
        SIAM Review, 1993




                                   
Polyblock Algorithm
    ●
        Current research in optimization is mainly to have 
        that third event ­ Solving non­convex optimization 
        efficiently. Although solving convex optimization 
        problems is easy and non­convex optimization 
        problems is hard, but a variety of approaches have 
        been proposed to solve non­convex optimization 
        problems.



                                  
Polyblock Algorithm
    ●
        In 2000, H. Tuy proposed an algorithm to solve 
        optimization problems involving d.i functions under 
        monotonic constraints.
    ●
        This algorithm (Polyblock Algorithm) was inspired 
        by the idea of Polyhedral Outer Approximation 
        Method for maximizing a quasi­convex function 
        over a convex set.


                                 
Polyblock Algorithm
    ●
        What is a polyblock? 
    ●
        Then what is the difference between a polyblock and 
        a polyhedron?
    ●
        What are its properties?
    ●
        How is polyblock algorithm implemented?




                                    
Polyblock Algorithm as in
[1]




               
Implementation of Polyblock
Algorithm
    ●
        Consider the following optimization problem:            
                        minimize    x1 + x2                                         
                        such that     (x1­3)  + 9(x2­3) ≥ 0                  
                                               3


                                           5x1 + 6x2 – 36 ≤ 0                     
                                           (x1,x2) ∊ [0,6]2




                                            
Implementation of Polyblock
Algorithm
    ●
        Feasible region of the problem




                                 
Implementation of Polyblock
Algorithm




              
Introduction


3 topics to be discussed:
1) Polyblock Algorithm for Monotonic Optimization
2) Network Utility Maximization
3) Internet Congestion Control Problem




                         
Network Utility
Maximization
    ●
        The framework of Network Utility Maximization 
        (NUM) has found many applications in network rate 
        allocation algorithms and Internet Congestion 
        Control Protocols.




                                
Network Utility
Maximization
    ●
        Problem: Consider a network with L links, each with 
        a fixed capacity cl bps, and S sources (i.e. end 
        users), each transmitting at the rate of xs bps. Each 
        source s uses the set L(s) of links in its path and has 
        a utility function Us(xs). Each link l is shared by a set 
        S(l) of sources. So, Network Utility Maximization is 
        basically the problem of maximizing the total utility 
        of the system over source rates subject to congestion 
        constraints for all links.
                                    
Network Utility
Maximization

    Mathematically,




                       
Network Utility
Maximization
    ●
        Concave Utilities ­ Follows from Law of 
        Diminishing Marginal Utilities. Convex 
        Optimization Problem.
    ●
        U(x) = log (1+x)

        U(x)




                                       x

                                 
NUM for Concave Utilities
    ●
        The problem of Network Utility Maximization in 
        case of concave utilities is essentially a convex 
        optimization problem which is solvable efficiently 
        and exactly.




                                  
Network Utility
Maximization
    ●
        Non­Concave Utilities – In multimedia applications 
        on Internet, the utilities are non­concave. Non­
        convex optimization problem. 
    ●
        U(x) =  (1 + e­ax+b) ­1 


        U(x)




                                       x
                                    
NUM for Non-Concave
Utilities
    ●
        The problem is a non­convex optimization problem. 
        Three ways have been suggested to solve it.
    ●
        In [3],  a 'self­regulation' heuristic is proposed, 
        however it converges only to a sub­optimal solution.
    ●
        In [4], a set of sufficient and necessary conditions is 
        presented under which the canonical distributed 
        algorithm converges to a global optimal solution. 
        However, these conditions may not hold in most 
        cases.
                                   
NUM for Non-Concave
Utilities
    ●
        In [2], Using a family of convex SDP relaxations 
        based on the sum­of­squares method and 
        Positivestellensatz Theorem in real algebraic 
        geometry, a centralized computational method to 
        bound the total network utility in polynomial time is 
        proposed.
    ●
        This is effectively a centralized method to compute 
        the global optimum when the utilities can be 
        transformed into polynomial utilities.
                                  
NUM for Non-Concave
Utilities
    ●
        In summary, currently there is no theoretically 
        polynomial­time algorithm (distributed or 
        centralised) known for non­concave utility 
        maximization.
    ●
        We worked to find ways to convexify the above 
        problem. 




                                  
Idea and motivation
    ●
        The set may not be a convex set but if it can be 
        broken into a constant number of convex sets, we 
        can solve the problem in polynomial time.




                                 
Idea and motivation




              
Idea and motivation




              
Idea and motivation




              
Motivation
    ●
        By this method, we can solve NUM problem in 
        polynomial time. NUM finds applications in 
        network rate allocation algorithms and Internet 
        Congestion Control Protocol.




                                  
Introduction


Not so much related topics:
1) Polyblock Algorithm for Monotonic Optimization
2) Network Utility Maximization
3) Internet Congestion Control Problem




                         
Internet Congestion Control
    ●
        Internet relies on congestion control implemented in 
        the end­systems to prevent offered load exceeding 
        network capacity, as well as allocate network 
        resources to different users and applications.
    ●
        In the past, the applications (email, file transfer) had 
        concave utilities (i.e were elastic). As number of 
        multimedia applications are increasing, there are 
        various talks on different congestion controls.

                                    
Internet Congestion Control
    ●
        In [5], It has been argued that fairness congestion 
        control does not maximize the network's utility. 
        Infact, Admission control is shown to be better 
        control (in terms of both elastic and inelastic 
        utilities) than Fair Congestion Control in a 
        simplified case. 
    ●
        Let α be the desired rate of inelastic flows, m be the 
        number of inelastic flows and n be the number of 
        elastic flows. 
                                   
Fair Congestion Control
    ●
        Perform TCP­friendly congestion control. We model 
        it as the same fair congestion control as adopted for 
        elastic flows, with a slight difference. When the fair 
        share is smaller than α, then the fair share is used, 
        but when the fair share is greater than α, the 
        inelastic flow would still consume α.




                                  
Admission Control
    ●
        Perform admission control but no congestion control 
        once admitted. Assume the network already has n 
        elastic flows and m inelastic flows, a new inelastic 
        flow is admitted iff nε + (m­1)α <=1
    ●
        Here ε represents the minimum rate admission 
        control scheme tries to leave for elastic traffic. 
        Depending upon α, we can have two cases:


                                    
Aggressive Admission
Control
    ●
        ε <<< α – The arriving flow is admitted as long as it 
        is possible to allocate to it the desired rate of α, even 
        if this means all elastic flows have to run at their 
        minimum rate of ε.
    ●
        So, an inelastic flow is admitted iff (m+1)α ≤ 1 and 
        an elastic flow is always admitted.




                                    
Fair Admission Control
    ●
        ε = α – The arriving flow is admitted as long as its 
        desired rate is no greater than the prevailing fair 
        share for each elastic flow.
    ●
        So, an inelastic flow is admitted iff (m+n+1)α ≤ 1 
        and an elastic flow is always admitted.
    ●
        In [5], it is proved that Fair Admission control is 
        better than both Aggressive Admission contol and 
        Fair Congestion Control.

                                  
Idea
    ●
        Solving the optimization problem using the 
        polyblock algorithm would help us to prove (or 
        disprove) that admission control is better than fair 
        congestion control.
    ●
        Status: Coding to check it under progress.




                                   
Thank You...........




                 
Bibliography
    ●
        [1] H. Tuy, ”Monotonic Optimization: Problems and Solution Approaches”, 
        SIAM Journal on Optimization, 11:2(2000), 464­494
    ●
        [2]  M. Fazel, M. Chiang, ”Network Utility Maximization With Nonconcave 
        Utilities Using Sum­of­Squares Method”, Proc. IEEE CDC, December 2005
    ●
        [3] J.W.Lee, R.R. Mazumdar, N. Shroff, ”Non­convex optimization and rate 
        control for multi­class services in the Internet”, Proc. IEEE Infocom, March 
        2004 
    ●
        [4] M. Chiang, S. Zhang, P. Hande, ”Distributed rate allocation for inelastic 
        flows: Optimization framework, optimality conditions, and optimal 
        algorithms”, Proc. IEEE Infocom, March 2005 
    ●
        [5] D. M. Chiu, A. ­S. W. Tam, ”Fairness of traffic controls for inelastic 
        flows in the Internet”, Comput. Netw. (2007), doi:10.1016/j.comnet.
        2006.12.2006                         

Más contenido relacionado

La actualidad más candente

Graphs in Data Structure
 Graphs in Data Structure Graphs in Data Structure
Graphs in Data Structurehafsa komal
 
Artificial Neural Networks Lect7: Neural networks based on competition
Artificial Neural Networks Lect7: Neural networks based on competitionArtificial Neural Networks Lect7: Neural networks based on competition
Artificial Neural Networks Lect7: Neural networks based on competitionMohammed Bennamoun
 
Optimization in Deep Learning
Optimization in Deep LearningOptimization in Deep Learning
Optimization in Deep LearningYan Xu
 
Batch normalization presentation
Batch normalization presentationBatch normalization presentation
Batch normalization presentationOwin Will
 
Software Architecture for Robotics
Software Architecture for RoboticsSoftware Architecture for Robotics
Software Architecture for RoboticsLorran Pegoretti
 
Graphical Models In Python | Edureka
Graphical Models In Python | EdurekaGraphical Models In Python | Edureka
Graphical Models In Python | EdurekaEdureka!
 
Machine learning with ADA Boost
Machine learning with ADA BoostMachine learning with ADA Boost
Machine learning with ADA BoostAman Patel
 
Poster - Convolutional Neural Networks for Real-time Road Sign Detection-V3Mr...
Poster - Convolutional Neural Networks for Real-time Road Sign Detection-V3Mr...Poster - Convolutional Neural Networks for Real-time Road Sign Detection-V3Mr...
Poster - Convolutional Neural Networks for Real-time Road Sign Detection-V3Mr...Guangrui Liu
 
Parallel sorting algorithm
Parallel sorting algorithmParallel sorting algorithm
Parallel sorting algorithmRicha Kumari
 
Basic Traversal and Search Techniques
Basic Traversal and Search TechniquesBasic Traversal and Search Techniques
Basic Traversal and Search TechniquesSVijaylakshmi
 
Training algorithms for Neural Networks
Training algorithms for Neural NetworksTraining algorithms for Neural Networks
Training algorithms for Neural NetworksMrinmoy Majumder
 
Particle swarm optimization
Particle swarm optimizationParticle swarm optimization
Particle swarm optimizationMahesh Tibrewal
 

La actualidad más candente (20)

Graphs in Data Structure
 Graphs in Data Structure Graphs in Data Structure
Graphs in Data Structure
 
Artificial Neural Networks Lect7: Neural networks based on competition
Artificial Neural Networks Lect7: Neural networks based on competitionArtificial Neural Networks Lect7: Neural networks based on competition
Artificial Neural Networks Lect7: Neural networks based on competition
 
Matching problem
Matching problemMatching problem
Matching problem
 
The Floyd–Warshall algorithm
The Floyd–Warshall algorithmThe Floyd–Warshall algorithm
The Floyd–Warshall algorithm
 
Optimization in Deep Learning
Optimization in Deep LearningOptimization in Deep Learning
Optimization in Deep Learning
 
Batch normalization presentation
Batch normalization presentationBatch normalization presentation
Batch normalization presentation
 
Asymptotic notation
Asymptotic notationAsymptotic notation
Asymptotic notation
 
Software Architecture for Robotics
Software Architecture for RoboticsSoftware Architecture for Robotics
Software Architecture for Robotics
 
Lasso regression
Lasso regressionLasso regression
Lasso regression
 
Graphical Models In Python | Edureka
Graphical Models In Python | EdurekaGraphical Models In Python | Edureka
Graphical Models In Python | Edureka
 
Introduction to Complex Networks
Introduction to Complex NetworksIntroduction to Complex Networks
Introduction to Complex Networks
 
Bagging.pptx
Bagging.pptxBagging.pptx
Bagging.pptx
 
Machine learning with ADA Boost
Machine learning with ADA BoostMachine learning with ADA Boost
Machine learning with ADA Boost
 
Poster - Convolutional Neural Networks for Real-time Road Sign Detection-V3Mr...
Poster - Convolutional Neural Networks for Real-time Road Sign Detection-V3Mr...Poster - Convolutional Neural Networks for Real-time Road Sign Detection-V3Mr...
Poster - Convolutional Neural Networks for Real-time Road Sign Detection-V3Mr...
 
Parallel sorting algorithm
Parallel sorting algorithmParallel sorting algorithm
Parallel sorting algorithm
 
R-CNN
R-CNNR-CNN
R-CNN
 
Basic Traversal and Search Techniques
Basic Traversal and Search TechniquesBasic Traversal and Search Techniques
Basic Traversal and Search Techniques
 
Training algorithms for Neural Networks
Training algorithms for Neural NetworksTraining algorithms for Neural Networks
Training algorithms for Neural Networks
 
Particle swarm optimization
Particle swarm optimizationParticle swarm optimization
Particle swarm optimization
 
Defuzzification
DefuzzificationDefuzzification
Defuzzification
 

Similar a Non-convex Optimization in Networks

Non concave network utility maximization - A distributed optimization approach
Non concave network utility maximization - A distributed optimization approachNon concave network utility maximization - A distributed optimization approach
Non concave network utility maximization - A distributed optimization approachWasif Hafeez
 
Introduction to algorithmic aspect of auction theory
Introduction to algorithmic aspect of auction theoryIntroduction to algorithmic aspect of auction theory
Introduction to algorithmic aspect of auction theoryAbner Chih Yi Huang
 
Green Optical Networks with Signal Quality Guarantee
Green Optical Networks with Signal Quality Guarantee Green Optical Networks with Signal Quality Guarantee
Green Optical Networks with Signal Quality Guarantee Maria Stylianou
 
Parallel Patterns for Window-based Stateful Operators on Data Streams: an Alg...
Parallel Patterns for Window-based Stateful Operators on Data Streams: an Alg...Parallel Patterns for Window-based Stateful Operators on Data Streams: an Alg...
Parallel Patterns for Window-based Stateful Operators on Data Streams: an Alg...Tiziano De Matteis
 
Quantum Computing: Timing is Everything
Quantum Computing: Timing is EverythingQuantum Computing: Timing is Everything
Quantum Computing: Timing is Everythinginside-BigData.com
 
Quantization and Training of Neural Networks for Efficient Integer-Arithmetic...
Quantization and Training of Neural Networks for Efficient Integer-Arithmetic...Quantization and Training of Neural Networks for Efficient Integer-Arithmetic...
Quantization and Training of Neural Networks for Efficient Integer-Arithmetic...Ryo Takahashi
 
The Traveling Salesman Problem: A Neural Network Perspective
The Traveling Salesman Problem: A Neural Network PerspectiveThe Traveling Salesman Problem: A Neural Network Perspective
The Traveling Salesman Problem: A Neural Network Perspectivemustafa sarac
 
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...Cemal Ardil
 
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks.pptx
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks.pptxEfficientNet: Rethinking Model Scaling for Convolutional Neural Networks.pptx
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks.pptxssuser2624f71
 
A White Paper On Neural Network Quantization
A White Paper On Neural Network QuantizationA White Paper On Neural Network Quantization
A White Paper On Neural Network QuantizationApril Knyff
 
PR-108: MobileNetV2: Inverted Residuals and Linear Bottlenecks
PR-108: MobileNetV2: Inverted Residuals and Linear BottlenecksPR-108: MobileNetV2: Inverted Residuals and Linear Bottlenecks
PR-108: MobileNetV2: Inverted Residuals and Linear BottlenecksJinwon Lee
 
Cosenza_MuleSoft_Meetup_1st.pptx
Cosenza_MuleSoft_Meetup_1st.pptxCosenza_MuleSoft_Meetup_1st.pptx
Cosenza_MuleSoft_Meetup_1st.pptxAndreaCapolei2
 
Technical Trends_Study of Quantum
Technical Trends_Study of QuantumTechnical Trends_Study of Quantum
Technical Trends_Study of QuantumHardik Gohel
 
Hands on machine learning with scikit-learn and tensor flow by ahmed yousry
Hands on machine learning with scikit-learn and tensor flow by ahmed yousryHands on machine learning with scikit-learn and tensor flow by ahmed yousry
Hands on machine learning with scikit-learn and tensor flow by ahmed yousryAhmed Yousry
 

Similar a Non-convex Optimization in Networks (20)

Non concave network utility maximization - A distributed optimization approach
Non concave network utility maximization - A distributed optimization approachNon concave network utility maximization - A distributed optimization approach
Non concave network utility maximization - A distributed optimization approach
 
Introduction to algorithmic aspect of auction theory
Introduction to algorithmic aspect of auction theoryIntroduction to algorithmic aspect of auction theory
Introduction to algorithmic aspect of auction theory
 
Green Optical Networks with Signal Quality Guarantee
Green Optical Networks with Signal Quality Guarantee Green Optical Networks with Signal Quality Guarantee
Green Optical Networks with Signal Quality Guarantee
 
Parallel Patterns for Window-based Stateful Operators on Data Streams: an Alg...
Parallel Patterns for Window-based Stateful Operators on Data Streams: an Alg...Parallel Patterns for Window-based Stateful Operators on Data Streams: an Alg...
Parallel Patterns for Window-based Stateful Operators on Data Streams: an Alg...
 
Introduction to optimization Problems
Introduction to optimization ProblemsIntroduction to optimization Problems
Introduction to optimization Problems
 
Quantum Computing: Timing is Everything
Quantum Computing: Timing is EverythingQuantum Computing: Timing is Everything
Quantum Computing: Timing is Everything
 
Parallel Algorithms
Parallel AlgorithmsParallel Algorithms
Parallel Algorithms
 
Quantization and Training of Neural Networks for Efficient Integer-Arithmetic...
Quantization and Training of Neural Networks for Efficient Integer-Arithmetic...Quantization and Training of Neural Networks for Efficient Integer-Arithmetic...
Quantization and Training of Neural Networks for Efficient Integer-Arithmetic...
 
The Traveling Salesman Problem: A Neural Network Perspective
The Traveling Salesman Problem: A Neural Network PerspectiveThe Traveling Salesman Problem: A Neural Network Perspective
The Traveling Salesman Problem: A Neural Network Perspective
 
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...
A weighted-sum-technique-for-the-joint-optimization-of-performance-and-power-...
 
N03430990106
N03430990106N03430990106
N03430990106
 
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks.pptx
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks.pptxEfficientNet: Rethinking Model Scaling for Convolutional Neural Networks.pptx
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks.pptx
 
Practical ML
Practical MLPractical ML
Practical ML
 
A White Paper On Neural Network Quantization
A White Paper On Neural Network QuantizationA White Paper On Neural Network Quantization
A White Paper On Neural Network Quantization
 
PR-108: MobileNetV2: Inverted Residuals and Linear Bottlenecks
PR-108: MobileNetV2: Inverted Residuals and Linear BottlenecksPR-108: MobileNetV2: Inverted Residuals and Linear Bottlenecks
PR-108: MobileNetV2: Inverted Residuals and Linear Bottlenecks
 
Cosenza_MuleSoft_Meetup_1st.pptx
Cosenza_MuleSoft_Meetup_1st.pptxCosenza_MuleSoft_Meetup_1st.pptx
Cosenza_MuleSoft_Meetup_1st.pptx
 
Technical Trends_Study of Quantum
Technical Trends_Study of QuantumTechnical Trends_Study of Quantum
Technical Trends_Study of Quantum
 
Hands on machine learning with scikit-learn and tensor flow by ahmed yousry
Hands on machine learning with scikit-learn and tensor flow by ahmed yousryHands on machine learning with scikit-learn and tensor flow by ahmed yousry
Hands on machine learning with scikit-learn and tensor flow by ahmed yousry
 
Power systemsilablri
Power systemsilablriPower systemsilablri
Power systemsilablri
 
1801.06434
1801.064341801.06434
1801.06434
 

Más de Pratik Poddar

Guide to wall street quant jobs for IITians
Guide to wall street quant jobs for IITiansGuide to wall street quant jobs for IITians
Guide to wall street quant jobs for IITiansPratik Poddar
 
Art of Puzzle Solving
Art of Puzzle SolvingArt of Puzzle Solving
Art of Puzzle SolvingPratik Poddar
 
Clipr rodinhood openhouse
Clipr rodinhood openhouseClipr rodinhood openhouse
Clipr rodinhood openhousePratik Poddar
 
What is Cryptography?
What is Cryptography?What is Cryptography?
What is Cryptography?Pratik Poddar
 
Audio Watermarking and Steganography
Audio Watermarking and SteganographyAudio Watermarking and Steganography
Audio Watermarking and SteganographyPratik Poddar
 
Identity Based Encryption
Identity Based EncryptionIdentity Based Encryption
Identity Based EncryptionPratik Poddar
 
Security Attacks on RSA
Security Attacks on RSASecurity Attacks on RSA
Security Attacks on RSAPratik Poddar
 
Ant Colony Optimization
Ant Colony OptimizationAnt Colony Optimization
Ant Colony OptimizationPratik Poddar
 

Más de Pratik Poddar (10)

Guide to wall street quant jobs for IITians
Guide to wall street quant jobs for IITiansGuide to wall street quant jobs for IITians
Guide to wall street quant jobs for IITians
 
Art of Puzzle Solving
Art of Puzzle SolvingArt of Puzzle Solving
Art of Puzzle Solving
 
Clipr Introduction
Clipr IntroductionClipr Introduction
Clipr Introduction
 
Clipr rodinhood openhouse
Clipr rodinhood openhouseClipr rodinhood openhouse
Clipr rodinhood openhouse
 
Grad School101
Grad School101Grad School101
Grad School101
 
What is Cryptography?
What is Cryptography?What is Cryptography?
What is Cryptography?
 
Audio Watermarking and Steganography
Audio Watermarking and SteganographyAudio Watermarking and Steganography
Audio Watermarking and Steganography
 
Identity Based Encryption
Identity Based EncryptionIdentity Based Encryption
Identity Based Encryption
 
Security Attacks on RSA
Security Attacks on RSASecurity Attacks on RSA
Security Attacks on RSA
 
Ant Colony Optimization
Ant Colony OptimizationAnt Colony Optimization
Ant Colony Optimization
 

Último

[BuildWithAI] Introduction to Gemini.pdf
[BuildWithAI] Introduction to Gemini.pdf[BuildWithAI] Introduction to Gemini.pdf
[BuildWithAI] Introduction to Gemini.pdfSandro Moreira
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century educationjfdjdjcjdnsjd
 
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...Jeffrey Haguewood
 
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemkeProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemkeProduct Anonymous
 
Platformless Horizons for Digital Adaptability
Platformless Horizons for Digital AdaptabilityPlatformless Horizons for Digital Adaptability
Platformless Horizons for Digital AdaptabilityWSO2
 
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingRepurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingEdi Saputra
 
Artificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyArtificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyKhushali Kathiriya
 
FWD Group - Insurer Innovation Award 2024
FWD Group - Insurer Innovation Award 2024FWD Group - Insurer Innovation Award 2024
FWD Group - Insurer Innovation Award 2024The Digital Insurer
 
Polkadot JAM Slides - Token2049 - By Dr. Gavin Wood
Polkadot JAM Slides - Token2049 - By Dr. Gavin WoodPolkadot JAM Slides - Token2049 - By Dr. Gavin Wood
Polkadot JAM Slides - Token2049 - By Dr. Gavin WoodJuan lago vázquez
 
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdfRising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdfOrbitshub
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FMESafe Software
 
MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MIND CTI
 
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...DianaGray10
 
Introduction to Multilingual Retrieval Augmented Generation (RAG)
Introduction to Multilingual Retrieval Augmented Generation (RAG)Introduction to Multilingual Retrieval Augmented Generation (RAG)
Introduction to Multilingual Retrieval Augmented Generation (RAG)Zilliz
 
Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherRemote DBA Services
 
Mcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot Model
Mcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot ModelMcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot Model
Mcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot ModelDeepika Singh
 
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...Angeliki Cooney
 
Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...
Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...
Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...apidays
 

Último (20)

[BuildWithAI] Introduction to Gemini.pdf
[BuildWithAI] Introduction to Gemini.pdf[BuildWithAI] Introduction to Gemini.pdf
[BuildWithAI] Introduction to Gemini.pdf
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century education
 
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
 
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemkeProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
ProductAnonymous-April2024-WinProductDiscovery-MelissaKlemke
 
Platformless Horizons for Digital Adaptability
Platformless Horizons for Digital AdaptabilityPlatformless Horizons for Digital Adaptability
Platformless Horizons for Digital Adaptability
 
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingRepurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
 
Artificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyArtificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : Uncertainty
 
FWD Group - Insurer Innovation Award 2024
FWD Group - Insurer Innovation Award 2024FWD Group - Insurer Innovation Award 2024
FWD Group - Insurer Innovation Award 2024
 
Polkadot JAM Slides - Token2049 - By Dr. Gavin Wood
Polkadot JAM Slides - Token2049 - By Dr. Gavin WoodPolkadot JAM Slides - Token2049 - By Dr. Gavin Wood
Polkadot JAM Slides - Token2049 - By Dr. Gavin Wood
 
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdfRising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
 
MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024
 
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
 
Introduction to Multilingual Retrieval Augmented Generation (RAG)
Introduction to Multilingual Retrieval Augmented Generation (RAG)Introduction to Multilingual Retrieval Augmented Generation (RAG)
Introduction to Multilingual Retrieval Augmented Generation (RAG)
 
Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a Fresher
 
Mcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot Model
Mcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot ModelMcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot Model
Mcleodganj Call Girls 🥰 8617370543 Service Offer VIP Hot Model
 
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...
Biography Of Angeliki Cooney | Senior Vice President Life Sciences | Albany, ...
 
Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...
Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...
Apidays New York 2024 - Passkeys: Developing APIs to enable passwordless auth...
 
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
 
Understanding the FAA Part 107 License ..
Understanding the FAA Part 107 License ..Understanding the FAA Part 107 License ..
Understanding the FAA Part 107 License ..
 

Non-convex Optimization in Networks

  • 1. Summer 2008 Internship Report Advisor: Prof. Angela Y. Zhang The Chinese University of Hong Kong, Hong Kong Student: Pratik Poddar Indian Institute of Technology Bombay, India Topic: Non­Convex Optimization  Problems in Networks    
  • 3. Basics of Optimization ● Standard Optimization problem ● Linear Optimization problem ● Convex Optimization problem ● Monotonic Optimization problem    
  • 4. Polyblock Algorithm ● We have had two major events in the history of  optimization theory. ●  The first was linear programming and simplex  method in late 1940s­ early 1950s. ● The second was convex optimization and interior  point method in late 1980s­ early 1990s.     
  • 5. Polyblock Algorithm ● Convex optimization problems are known to be  solved, very reliably and efficiently. ● "..in fact, the great watershed in optimization isn't  between linearity and nonlinearity, but convexity  and nonconvexity" ­ R. Tyrrell Rockafellar, in   SIAM Review, 1993    
  • 6. Polyblock Algorithm ● Current research in optimization is mainly to have  that third event ­ Solving non­convex optimization  efficiently. Although solving convex optimization  problems is easy and non­convex optimization  problems is hard, but a variety of approaches have  been proposed to solve non­convex optimization  problems.    
  • 7. Polyblock Algorithm ● In 2000, H. Tuy proposed an algorithm to solve  optimization problems involving d.i functions under  monotonic constraints. ● This algorithm (Polyblock Algorithm) was inspired  by the idea of Polyhedral Outer Approximation  Method for maximizing a quasi­convex function  over a convex set.    
  • 8. Polyblock Algorithm ● What is a polyblock?  ● Then what is the difference between a polyblock and  a polyhedron? ● What are its properties? ● How is polyblock algorithm implemented?    
  • 9. Polyblock Algorithm as in [1]    
  • 10. Implementation of Polyblock Algorithm ● Consider the following optimization problem:                             minimize    x1 + x2                                                          such that     (x1­3)  + 9(x2­3) ≥ 0                   3                                    5x1 + 6x2 – 36 ≤ 0                                                         (x1,x2) ∊ [0,6]2    
  • 11. Implementation of Polyblock Algorithm ● Feasible region of the problem    
  • 14. Network Utility Maximization ● The framework of Network Utility Maximization  (NUM) has found many applications in network rate  allocation algorithms and Internet Congestion  Control Protocols.    
  • 15. Network Utility Maximization ● Problem: Consider a network with L links, each with  a fixed capacity cl bps, and S sources (i.e. end  users), each transmitting at the rate of xs bps. Each  source s uses the set L(s) of links in its path and has  a utility function Us(xs). Each link l is shared by a set  S(l) of sources. So, Network Utility Maximization is  basically the problem of maximizing the total utility  of the system over source rates subject to congestion  constraints for all links.    
  • 16. Network Utility Maximization Mathematically,    
  • 17. Network Utility Maximization ● Concave Utilities ­ Follows from Law of  Diminishing Marginal Utilities. Convex  Optimization Problem. ● U(x) = log (1+x) U(x) x    
  • 18. NUM for Concave Utilities ● The problem of Network Utility Maximization in  case of concave utilities is essentially a convex  optimization problem which is solvable efficiently  and exactly.    
  • 19. Network Utility Maximization ● Non­Concave Utilities – In multimedia applications  on Internet, the utilities are non­concave. Non­ convex optimization problem.  ● U(x) =  (1 + e­ax+b) ­1  U(x) x    
  • 20. NUM for Non-Concave Utilities ● The problem is a non­convex optimization problem.  Three ways have been suggested to solve it. ● In [3],  a 'self­regulation' heuristic is proposed,  however it converges only to a sub­optimal solution. ● In [4], a set of sufficient and necessary conditions is  presented under which the canonical distributed  algorithm converges to a global optimal solution.  However, these conditions may not hold in most  cases.    
  • 21. NUM for Non-Concave Utilities ● In [2], Using a family of convex SDP relaxations  based on the sum­of­squares method and  Positivestellensatz Theorem in real algebraic  geometry, a centralized computational method to  bound the total network utility in polynomial time is  proposed. ● This is effectively a centralized method to compute  the global optimum when the utilities can be  transformed into polynomial utilities.    
  • 22. NUM for Non-Concave Utilities ● In summary, currently there is no theoretically  polynomial­time algorithm (distributed or  centralised) known for non­concave utility  maximization. ● We worked to find ways to convexify the above  problem.     
  • 23. Idea and motivation ● The set may not be a convex set but if it can be  broken into a constant number of convex sets, we  can solve the problem in polynomial time.    
  • 27. Motivation ● By this method, we can solve NUM problem in  polynomial time. NUM finds applications in  network rate allocation algorithms and Internet  Congestion Control Protocol.    
  • 29. Internet Congestion Control ● Internet relies on congestion control implemented in  the end­systems to prevent offered load exceeding  network capacity, as well as allocate network  resources to different users and applications. ● In the past, the applications (email, file transfer) had  concave utilities (i.e were elastic). As number of  multimedia applications are increasing, there are  various talks on different congestion controls.    
  • 30. Internet Congestion Control ● In [5], It has been argued that fairness congestion  control does not maximize the network's utility.  Infact, Admission control is shown to be better  control (in terms of both elastic and inelastic  utilities) than Fair Congestion Control in a  simplified case.  ● Let α be the desired rate of inelastic flows, m be the  number of inelastic flows and n be the number of  elastic flows.     
  • 31. Fair Congestion Control ● Perform TCP­friendly congestion control. We model  it as the same fair congestion control as adopted for  elastic flows, with a slight difference. When the fair  share is smaller than α, then the fair share is used,  but when the fair share is greater than α, the  inelastic flow would still consume α.    
  • 32. Admission Control ● Perform admission control but no congestion control  once admitted. Assume the network already has n  elastic flows and m inelastic flows, a new inelastic  flow is admitted iff nε + (m­1)α <=1 ● Here ε represents the minimum rate admission  control scheme tries to leave for elastic traffic.  Depending upon α, we can have two cases:    
  • 33. Aggressive Admission Control ● ε <<< α – The arriving flow is admitted as long as it  is possible to allocate to it the desired rate of α, even  if this means all elastic flows have to run at their  minimum rate of ε. ● So, an inelastic flow is admitted iff (m+1)α ≤ 1 and  an elastic flow is always admitted.    
  • 34. Fair Admission Control ● ε = α – The arriving flow is admitted as long as its  desired rate is no greater than the prevailing fair  share for each elastic flow. ● So, an inelastic flow is admitted iff (m+n+1)α ≤ 1  and an elastic flow is always admitted. ● In [5], it is proved that Fair Admission control is  better than both Aggressive Admission contol and  Fair Congestion Control.    
  • 35. Idea ● Solving the optimization problem using the  polyblock algorithm would help us to prove (or  disprove) that admission control is better than fair  congestion control. ● Status: Coding to check it under progress.    
  • 37. Bibliography ● [1] H. Tuy, ”Monotonic Optimization: Problems and Solution Approaches”,  SIAM Journal on Optimization, 11:2(2000), 464­494 ● [2]  M. Fazel, M. Chiang, ”Network Utility Maximization With Nonconcave  Utilities Using Sum­of­Squares Method”, Proc. IEEE CDC, December 2005 ● [3] J.W.Lee, R.R. Mazumdar, N. Shroff, ”Non­convex optimization and rate  control for multi­class services in the Internet”, Proc. IEEE Infocom, March  2004  ● [4] M. Chiang, S. Zhang, P. Hande, ”Distributed rate allocation for inelastic  flows: Optimization framework, optimality conditions, and optimal  algorithms”, Proc. IEEE Infocom, March 2005  ● [5] D. M. Chiu, A. ­S. W. Tam, ”Fairness of traffic controls for inelastic  flows in the Internet”, Comput. Netw. (2007), doi:10.1016/j.comnet.   2006.12.2006