Optimal Resource Allocation in Coordinated Multi-Cell Systems Emil Björnson Post-Doc Alcatel-Lucent Chair on Flexible Radio, Supélec & Signal Processing Lab, KTH Royal Institute of Technology Seminar 2012-10-11 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 1 Biography: Emil Björnson • 1983: Born in Malmö, Sweden • 2007: Master of Science in Engineering Mathematics, Lund University, Sweden • 2011: PhD in telecommunications, KTH, Stockholm, Sweden • 2012: Recipient of International Postdoc Grant from Sweden. Work with M. Debbah at Supélec for next 2 years. Optimal Resource Allocation in Coordinated Multi-Cell Systems Monograph Tutorial by E. Björnson and E. Jorswieck Submitted to FnT Communications and Information Theory 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 2 Outline • Introduction - Multi-Cell Structure, System Model, Performance Measure • Problem Formulation - Resource Allocation: Multi-Objective Optimization Problem • Subjective Resource Allocation - Utility Functions, Different Computational Complexity • Structural Insights - Beamforming Parametrization • Some Practical Extensions - Handling Non-Idealities in Practical Systems 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 3 Introduction 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 4 Introduction • Problem Formulation (vaguely): - Transfer Information Wirelessly • Downlink Coordinated Multi-Cell System - Many Transmitting Base Stations (BSs) Many Receiving Users Sharing a Common Frequency Band Limiting Factor: Inter-User Interference • Multi-Antenna Transmission - Beamforming: Spatially Directed Signals - Can Serve Multiple Users (Simultaneously) 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 5 Introduction: Basic Multi-Cell Structure • One Base Station per Cell - Adjacent Base Stations Coordinate Interference - Some Users Served by Multiple Base Stations • Dynamic Cooperation Clusters - Inner Circle: Serve Users with Data - Outer Circle: Avoid Interference - Outside Circles: Negligible Impact (Impractical to Coordinate) 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 6 Example: Ideal Joint Transmission • All Base Stations Serve All Users Jointly 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 7 Example: Wyner Model • Abstraction: User receives signals from own and neighboring base stations • One or two dimensional versions • Joint transmission or coordination between cells 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 8 Example: Coordinated Beamforming • One base station serves each user • Interference coordination across cells 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 9 Example: Cognitive Radio Other Examples Spectrum Sharing between Operators Physical Layer Security • Secondary System Borrows Spectrum of Primary System • Underlay: Interference Limits for Primary Users 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 10 Introduction: Resource Allocation • Problem Formulation (imprecise): - Select Beamforming to Maximize System Utility - Means: Allocate Power to Users and in Spatial Dimensions - Satisfy: Physical, Regulatory & Economic Constraints • Some Assumptions: - Linear Transmission and Reception - Perfect Synchronization (where needed) - Flat-fading Channels (e.g., using OFDM) - Perfect Channel State Information - Ideal Transceiver Hardware - Centralized Optimization 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH Will be relaxed 11 Introduction: Multi-Cell System Model • πΎπ Users: Channel Vector hπ to User π from All BSs • ππ Antennas at πth BS • π= 2012-10-11 π ππ Antennas in Total (dimension of hπ ) Emil Björnson, Post-Doc at SUPELEC and KTH 12 Introduction: Power Constraints • πΏ General Power Constraints: Weight Matrix (Positive semi-definite) Limit (Positive scalar) • Many Interpretations 2012-10-11 Protect Dynamic Range of Amplifiers Limit Radiated Power According to Regulations Control Interference to Certain Users Manage Energy Expenditure Emil Björnson, Post-Doc at SUPELEC and KTH 13 Introduction: User Performance Measure • Mean Square Error (MSE) - Difference: transmitted and received signal - Easy to Analyze - Far from User Perspective? • Bit/Symbol Error Rate (BER/SER) - Probability of Error (for given data rate) - Intuitive Interpretation - Complicated & ignores channel coding All improves with SINR: Signal Interf + Noise • Information Rate - Bits per ”channel use” - Mutual information: perfect and long coding - Still closest to reality? 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 14 Introduction: User Performance Measure • Generic Model - Any Function of Signal-to-Interference-and-Noise Ratio (SINR) - Continuous and Increasing Function - Can be MSE, BER/SER, Information Rate, etc. • Complicated Function - Depends on All Beamforming Vectors v1 , … , vπΎπ 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 15 Problem Formulation 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 16 Problem Formulation • General Formulation of Resource Allocation: • Multi-Objective Optimization Problem - Generally Impossible to Maximize For All Users! - Must Divide Power and Cause Inter-User Interference 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 17 Performance Region • Definition: Performance Region R - Contains All Feasible Care about user 1 Pareto Boundary Balance between users Part of interest: Pareto boundary 2 User Performance Region 2012-10-11 Cannot Improve for any user without degrading for other users Care about user 2 Emil Björnson, Post-Doc at SUPELEC and KTH 18 Performance Region (2) • Can it have any shape? • No! Can prove that: - Compact set - Simply connected (No holes) - Nice upper boundary Normal set Upper corner in region, everything inside region 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 19 Performance Region (3) • Some Possible Shapes Shape is Unknown User-Coupling Weak: Convex Strong: Concave 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 20 Performance Region (4) • Which Pareto Optimal Point to Choose? - Tradeoff: Aggregate Performance vs. Fairness Utilitarian point (Max sum performance) Egalitarian point (Max fairness) Single user point 2012-10-11 Performance Region No Objective Answer Only Subjective Answers Exist! Single user point Emil Björnson, Post-Doc at SUPELEC and KTH 21 Subjective Resource Allocation 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 22 Subjective Approach • System Designer Selects Utility Function f : R → R - Describing Subjective Preference - Increasing Function • Examples: Sum Performance: Proportional Fairness: Harmonic Mean: Max-Min Fairness: 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 23 Subjective Approach (2) • Gives Single-Objective Optimization Problem: • This is the Starting Point of Many Researchers - Although Selection of f is Inherently Subjective Affects the Solvability Pragmatic Approach Try to Select Utility Function to Enable Efficient Optimization 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 24 Subjective Approach (3) • Characterization of Optimization Problems • Main Categories of Resource Allocation - Convex: Polynomial Time Solution - Monotonic: Exponential Time Solution 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH Practically Solvable Approx. Needed 25 Subjective Approach (4) • When is the Problem Convex? - Most Problems are Non-Convex - Necessary: Search Space must be Particularly Limited • We will Classify Three Important Problems - The “Easy” Problem - Weighted Max-Min Fairness - Weighted Sum Performance 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 26 The “Easy” Problem • Given Any Point (π1 , … , ππΎπ ) - Find Beamforming v1 , … , vπΎπ that Attains this Point - Minimize Power • Convex Problem - Second-Order Cone or Semi-Definite Program - Global Solution in Polynomial Time – use CVX, Yalmip Total Power Constraints • M. Bengtsson, B. Ottersten, “Optimal Downlink Beamforming Using Semidefinite Optimization,” Proc. Allerton, 1999. • A. Wiesel, Y. Eldar, and S. Shamai, “Linear precoding via conic optimization for fixed MIMO receivers,” IEEE Trans. Signal Processing, 2006. Per-Antenna Constraints • W. Yu and T. Lan, “Transmitter optimization for the multi-antenna downlink with per-antenna power constraints,” IEEE Trans. Signal Process., 2007. General Constraints, Robustness • E. Björnson, G. Zheng, M. Bengtsson, B. Ottersten, “Robust Monotonic Optimization Framework for Multicell MISO Systems,” IEEE Transactions on Signal Processing, IEEE Trans. Signal Process., 2012. 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 27 Subjective Approach: Max-Min Fairness • How to Classify Weighted Max-Min Fairness? - Property: Solution makes π€π ππ the same for all π Solution is on this line Line in direction (π€1 , … , π€πΎπ ) 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 28 Subjective Approach: Max-Min Fairness • Simple Line-Search: Bisection - Iteratively Solving Convex Problems (i.e., Quasi-convex) 1. Find start interval 2. Solve the “easy” problem at midpoint 3. If feasible: Remove lower half Else: Remove upper half 4. Iterate Subproblem: Convex optimization Line-search: Linear convergence One dimension (independ. #users) 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 29 Subjective Approach: Max-Min Fairness • Classification of Weighted Max-Min Fairness: - Quasi-Convex Problem (belongs to convex class) • If Subjective Preference is Formulated in this Way - Resource Allocation Solvable in Polynomial Time 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 30 Subjective Approach: Sum Performance • How to Classify Weighted Sum Performance? - Geometrically: π€1 π1 + π€2 π2 = opt-value is a line Opt-value is unknown! - 2012-10-11 Distance from origin is unknown Line ο Hyperplane (dim: #user – 1) Harder than max-min fairness Provably NP-hard! Emil Björnson, Post-Doc at SUPELEC and KTH 31 Subjective Approach: Sum Performance • Classification of Weighted Sum Performance: - Monotonic Problem • If Subjective Preference is Formulated in this Way - Resource Allocation solvable in Exponential Time • Algorithm for Monotonic Optimization - Improve lower/upper bounds on optimum: - Continue until - Subproblem: Essentially Weighted Max-Min Fairness 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 32 Subjective Approach: Sum Performance Branch-Reduce-Bound (BRB) Algorithm - Global Convergence - Accuracy ε>0 in finitely many iterations - Exponential complexity only in #users (πΎπ ) - Polynomial complexity in other parameters (#antennas/constraints) 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 33 Pragmatic Resource Allocation • Recall: All Utility Functions are Subjective - Pragmatic Approach: Select to Enable Efficient Optimization • Bad Choice: Weighted Sum Performance - NP-hard: Exponential complexity (in #users) • Good Choice: Weighted Max-Min Fairness - Quasi-Convex: Polynomial complexity Pragmatic Resource Allocation Weighted Max-Min Fairness: Weights Enhance Throughput 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 34 Why is Weighted Sum Performance bad? • Some shortcomings - 2012-10-11 Law of Diminishing Marginal Utility not satisfied Not all Pareto Points are Attainable Weights have no Clear Interpretation Not Robust to Perturbations Emil Björnson, Post-Doc at SUPELEC and KTH 35 Structural Insights 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 36 Parametrization of Optimal Beamforming • Any Resource Allocation Problem Solved by - Original πΎπ π Complex → πΎπ + πΏ– 2 Positive Parameters - Priority of User π: ππ - Impact of Constraint π: ππ • Tradeoff: Maximize Signal - Minimize Interference - Not easy to find the best tradeoff 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 37 Parametrization of Optimal Beamforming • Geometric Interpretation: • Heuristic Parameter Selection - Known to Work Remarkably Well - Many Examples (since 1995): Transmit Wiener/MMSE filter, Regularized Zero-forcing, Signal-to-leakage beamforming, virtual SINR/MVDR beamforming, etc. 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 38 Some Practical Extensions 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 39 Robustness to Channel Uncertainty • Practical Systems Operate under Uncertainty - Due to Estimation, Feedback, Delays, etc. • Robustness to Uncertainty - Maximize Worst-Case Performance - Cannot be Robust to Any Error • Ellipsoidal Uncertainty Sets - Easily Incorporated in the Model - More Variables – Same Classifications - Definition: 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 40 Distributed Resource Allocation • Information and Functionality is Distributed - Local Channel Knowledge and Computational Resources - Only Limited Backhaul for Coordination • Distributed Approach - Decompose Optimization - Exchange Control Signals - Iterate Subproblems • Convergence to Optimal Solution? - At least for Convex Problems 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 41 Adapting to Transceiver Impairments • Physical Hardware is Non-Ideal - Phase Noise, IQ-imbalance, Non-Linearities, etc. - Non-Negligible Performance Degradation at High SNR • Model of Transmitter Distortion: - Additive Noise - Variance Scales with Signal Power • Same Classifications Hold under this Model - Enables Adaptation: Much Larger Tolerance for Impairments 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 42 Summary 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 43 Summary • Resource Allocation - Divide Power between Users and Spatial Directions - Solve a Multi-Objective Optimization Problem - Pareto Boundary: Set of Efficient Solutions • Subjective Utility Function - Selection has Fundamental Impact on Solvability Pragmatic Approach: Select to Enable Efficient Optimization Weighted Sum Performance: Not Solvable in Practice Weighted Max-Min Fairness: Polynomial Complexity • Parametrization of Optimal Beamforming • Extensions: Channel Uncertainty, Distributed Optimization, Transceiver Impairments 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 44 Main Reference • 250-page tutorial monograph – currently under review • Many more details - Other convex problems and general algorithms - More parametrizations and structural insights - Extensions: multi-cast, multi-carrier, multi-antenna users, etc. 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 45 Thank You for Listening! Questions? All Papers Available: http://flexible-radio.com/emil-bjornson 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 46 Backup Slides 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 47 Problem Classifications General Zero Forcing Single Antenna Sum Performance NP-hard Convex NP-hard Proportional Fairness NP-hard Convex Convex Harmonic Mean NP-hard Convex Convex Max-Min Fairness Quasi-Convex Quasi-Convex Quasi-Convex QoS/Easy Problem Convex Convex Linear • General: Any Utility and Constraints • Zero Forcing: Concave Utilities and ZF Constraints • Single Antenna: ππ = 1 and one BS Serves Each User 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 48 Monotonic Optimization • Branch-Reduce-Bound (BRB) Algorithm 1. 2. 3. 4. Cover region with a box Divide the box into two sub-boxes Remove parts with no solutions in Search for solutions to improve bounds (Based on Fairness-profile problem) 5. Continue with sub-box with largest value 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 49 Computation of Performance Regions • Performance Region is Generally Unknown - Compact and Normal - Perhaps Non-Convex • Generate 1: Vary Parameters in Parametrization • Generate 2: Maximize Sequence of Utilities f 2012-10-11 Emil Björnson, Post-Doc at SUPELEC and KTH 50