CrossLayer Resource Allocation in Multiaccess Wireless Network: the Problems and One Solution

1
Cross-Layer Resource Allocation in Multi-access
Wireless Networkthe Problems and One Solution

• Zhu Han
• Department of Electrical and Computer
  Engineering,
• University of Maryland, College Park.
• At Carleton University
• 07/10/2003

2
Outline

• Cross Layer Resource Allocation for Multi-users
• What causes the problems
• Why cross layer approaches
• How to formulate the problems
• Four possible categories of solutions
• Further performance improvement techniques
• Power Minimization under Throughput Management
  over Wireless Network
• Joint power control and adaptive modulation
• Fairness constraint
• Heuristic solutions

3
What causes the problems

• Time Varying Channels
• Fading, shadowing, path losses, DOA, and noise
  levels
• Dynamic allocation of resources such as powers
  and rates
• Multi-access Wireless Networks
• FDMA, TDMA, CDMA, SDMA, CSMA,and OFDMA.
• System allocates resources such as channels to
  different users to increase the overall system
  performance.
• Limited Bandwidths
• Cellular concept channel reuse
• Reduce Co-channel interferences by cooperation

4
Why cross layer approaches

• Traditional Resource Optimization is Designed in
  Layers.
• Application layer multimedia source coder
• Network layer base station assignment, handoff,
  routing
• MAC layer queuing, adaptive rate and coding, QoS
  
• Physical layer power control, antenna array
  processing
• Shortcomings
• Layers dont have knowledge with each other
  local optima
• Overhead associated between layers becomes
  precious.
• Advantages and Disadvantages of Cross Layer
  Approach
• Optimal from resource management point of view
• Suboptimal from implementation point of view
  complex, nonlinear, non-convex, hard to model,
  and hard to implement.

5
Examples for cross layer problems

• Joint Source Channel Coding
• Distortion Based Resource Allocation for
  Multi-access Networks
• Joint Power Control, Beamforming and Base Station
  Assignment
• Dynamic Cell Management
• Routing for Ad-hoc Wireless Network with Power
  Constraint
• Joint Power Control and Handoff Technique
• Adaptive Resource Allocation with Buffer or Delay
  Constraint
• Resource Management in Packet Access Systems
• Joint Power Control and Rate Adaptation
• OFDMA Channel Allocation with Throughput
  Constraint

6
Resources, constraints, and problems

• Resources (Parameters) ?
• Transmitted power, rate (source rate, channel
  rate, symbol rate), base station, channel,
  antenna weight vector....
• Constraints ?
• Maximal transmitted power
• Maximal delay
• Optimization Goals ?
• Overall throughput, overall transmitted power,
  average distortion, maximum outage rate, overall
  QoS, or multiple objectives.
• General Problem Formulation

7
Possible solution analysis

• Approximation and Simplification
• Make the problems to be linear or convex.
• Methods
• Lagrange multiplier
• Convex optimization
• Results
• Clean analytical results
• Fast converged algorithms.
• Advantages and Disadvantages
• Beautiful mathematics, claimed optima,
  reviewers favorite
• Performance is highly related to how good the
  approximation and simplification to the reality.
  (How far away from truth?)

8
Possible solution optimal control

• Constrained Optimization Problem
• Nonlinear programming
• Integer programming
• Advantages and Disadvantages
• Close to reality
• Fast, no iteration between BS and users is
  needed.
• Centralized control
• Multiple optima multiple initializations or
  annealing
• Full knowledge of channel conditions
• High complexity with large number of parameters
  and users
• Performance bound. Only fit the centralized
  system with a small number of users, such as a
  CDMA micro cell.
• Distributed Implementation Pricing, Simple Cases

9
Possible solution game theory

• Non-cooperative Game
• Individual mobile users dont have knowledge of
  other users conditions and cannot cooperate with
  others, they act selfishly to maximize their own
  performances in a distributed manner.
• Non-cooperative game deals largely with how
  rational and intelligent individuals interact
  with each other in an effort to achieve their own
  goals.
• Utility functions, Games, and Nash Equilibriums.
• Nash Equilibriums May Not Be Efficient for System
• Goal design meaningful utility functions and
  game rules, so that the system is balanced in the
  desired social optimal equilibrium.
• Pricing develop a mediator between the users
  interests and the system efficiency. demand and
  supply rule.
• Repeated Game Act by the rules to avoid future
  punishments.
• Cooperative Game Bargain. Like a company

10
Possible solution dynamic programming

• Optimization over Different Time.
• Dynamic programming make the optimal decisions
  over time, based on the distributions of the
  channels or the sources.
• Scheduling A Special Case of Dynamic Programming
• Tradeoff for the fairness (delay) and the system
  performance.
• Advantages and Disadvantages
• Users might not be optimized at a specific time.
  But their sacrifices of performances will
  increase the overall system performance and will
  be compensated back in the future.
• Distributions for decision in a multi-user case
  are hard to obtain,
• High computation complexity
• Only fit single user, simple channel condition
  case.

11
Pro and Con for different solutions
12
Further performance improvements

• Antenna Array Processing
• Multi-User Detection
• Space Time Processing
• MIMO System
• Adaptive BICM, LDPC Coding
• OFDM Channel Assignment
• Memo for the resource allocation over the crowded
  wireless networks

Find a New Continent, or Manage the Old Continent
Efficiently.
13
Introduction

• Power Minimization under Throughput Management
  over Wireless Networks with Antenna Diversity
• Time Varying Nature and Co-Channel Interferences
• Power control and Adaptive modulation
• Minimize the Overall Transmitted Power of
  Networks.
• No reduction for the overall network throughput
• Fairness users time average throughput is
  assured
• Heuristically Divide the Problem into Two
  Sub-problems at the User Level and at the System
  Level, Respectively.
• Water Filling Each Users Throughput in Time
  Domain and Allocating Network Throughput to
  Different Users Each Time.

14
Motivation

• Existing Works
• Pure power control Received SINR is larger than
  a fixed threshold. Dont consider channel
  variances. All users are the same. Cause
  unnecessary CCI in deep fading.
• Pure throughput maximization Maximize the
  capacity (throughput) without considering
  fairness.
• Only the user with best channel can transmit.
• Socialism Idea
• Allow users to sacrifice their performances in a
  short period, with the incentive that the overall
  network transmitted power can be reduced and the
  users temporary sacrifices will be paid back in
  a long term.

15
System model

• Multi-cell, One User per Cell, TDMA/FDMA, Uplink
• Antenna Diversity
• Maximal Ratio Combine and Selective Combine
• Received SINR
• Adaptive Modulation MQAM

16
Traditional power control

• Problem Formulation
• Problems
• A fixed and predefined targeted SINR threshold
  .
• Works perfectly in low SINR areas.
• Powers increase quickly when the threshold is
  higher.
• No feasible solution if the threshold is too
  high.
• A user with a bad channel causes too much CCI.
• Adapt the thresholds, according to channel
  conditions.

17
Proposed problem formulation

• Problem Formulation
• Difficulties
• Bilinear Matrix Inequality
• Involve complicated dynamic programming

18
Problem partition

• Problem Partition into the User Level and the
  System Level
• Users adapt and report acceptable throughput
  ranges, according to
• their transmission histories and current channel
  conditions.
• If less throughput now, more
• aggressive to transmit, higher
• range in the future, so that fairness
• is maintained. Channel conditions
• are also considered
• System determines the optimal
• throughput allocation to min P,
• within these throughput ranges.
• Two-user Example

19
User Throughput range algorithm

• Idea Credit System. Moving Throughput Window
• If assigned lower throughput, the user
  accumulates credit, so that he can aggressively
  transmit by providing higher throughput windows
  in the future when the channel becomes good.
• High throughput, credit is used up, so less
  aggressive and smaller throughput windows.
  Fairness is maintain.
• Change the Throughput Windows according to
  Channel Trends.
• Throughput smaller than that of the adjacent
  cells, still in bad channel condition and report
  lower throughput window, vice versa.
• Extreme Case Analysis
• Trapped in bad channels, users can provide lowest
  
• throughput range to wait for channels to become
  better.
• Proof to Guarantee Fairness

20
User Throughput range algorithm

• Initialization
• Iteration
• If all adjacent CCI cells,
• Else
• Feedback the acceptable throughput range back to
  BS
• If
• report
• Else report

21
System Throughput allocation algorithm 1

• Full Search Algorithm
• Adaptive Modulation
• Search all possible throughput combinations
  subject to constraints. Find the combination that
  minimizes the overall power.
• Iteration
• Powers are initialized by any feasible values.
• Antenna diversity
• Power update
• Throughput Range Update
• Update
• Too Complex, Only for Comparing Performances.

22
System Throughput allocation algorithm 2

• Fast Search Algorithm
• Find the gradient of Psum with respect to users
  target SINR.
• For the user with the largest gradient, Find the
  throughput that generates the lowest overall
  transmitted power subject to the constraints.
• When the throughput of the user with the largest
  gradient is changed, the throughput of the other
  users is modified in the order from the lower
  gradient to higher gradient to compensate the
  network throughput constraint TR.
• More throughput is allocated to the users with
  small gradients, and less throughput is assigned
  to the users with large gradients.
• Simple but may be sub-optimal.

23
System Throughput allocation algorithm 3

• Projected Gradient Method
• Nonlinear programming by assuming throughput is
  continuous
• Constrained projected gradient method
• Projected the continuous throughput to the
  discrete values.
• More complex than fast search, can find optimal
  solutions

24
Simulation results
25
Simulation results
26
Simulation results
27
Conclusions

• Cross Layer Resource Allocation for Multi-access
  Wireless Networks
• For each user, cross layer approach can increase
  the performance.
• For whole system, clearly managing different
  users resources can increase the system
  performance and reduce unnecessary CCI.
• The desired scheme should have easy,
  (sub-)optimal, and distributed implementation,
  without requiring too much information.
• Joint Power and Throughput Optimization
• Fairness for the services that the users have
  paid for.
• Water filling each users throughput in time
  domain and allocating the network throughput to
  different users each time.
• 7 dB gain for powers, 1.2 bit/s/Hz gain for
  spectrum efficiencies.

28
My Other Works

• Game Theory and Economy Approach
• Distortion Based CDMA
• Joint Source-Channel Coding
• Estimation
• Blind Estimation
• Bio Image Processing
• Time of Arrival
• Beamforming
• Tutorial Paper
• Channel Allocation for OFDMA

29
Self-advertisement

• Journal Papers
• Zhu Han and K.J.Ray Liu, "Joint Adaptive Link
  Quality and Power Management with Fairness
  Constraint over Wireless Networks", Submitted to
  IEEE Transactions on Vehicular Technology.
• Zhu Han and K.J.Ray Liu, "Power Minimization
  under Throughput Management over Wireless
  Networks with Antenna Diversity", Revision, IEEE
  Transactions on Wireless Communications.
• Zhu Han and K.J.Ray Liu, "Joint Power Control and
  Blind Beamforming over Wireless Networks A Cross
  Layer Approach", Submitted to Eurasip.
• Zhu Han and K.J.Ray Liu, "Non-Cooperative Power
  Control Game and Throughput Game over Wireless
  Networks", Submitted to IEEE Transactions on
  Communications.
• Zhu Han, Jane Wang and K.J.Ray Liu, "A Resource
  Allocation Framework with Credit System and User
  Autonomy Over Heterogeneous Wireless Networks ",
  in Preparation.
• Zhu Han, Farrokh Rashid-Farrokhi and K.J.Ray Liu,
  "A Tutorial for Cross Layer Resource Allocation
  in Wireless Networks Problems, Techniques and
  Solutions", in Preparation.
• Zhu Han, Andres Kwasinski, Mehdi Alasti, K.J.Ray
  Liu, and Nariman Farvardin, "Downlink Resource
  Allocation for Multi-cell CDMA Networks Based on
  Real-Time Dynamic Joint Source Channel Coding",
  in Preparation.
• Jane Wang, Zhu Han and K.J.Ray Liu, "MIMO-OFDM
  Channel Estimation via Probabilistic Data
  Association Based TOA Estimation", in
  Preparation.
• Jane Wang, Zhu Han and K.J.Ray Liu, "DCE-MRI
  Based Tumor Heterogeneity Characterization
  Simultaneous Estimation of Kinetic Parameters and
  Input Function", in Preparation.

30
Self-advertisement

• Conference Papers
• Zhu Han and K.J.Ray Liu, "Non-Cooperative Power
  Control Game and Throughput Game over Wireless
  Networks, submitted to Infocom 2004.
• Zhu Han, Andres Kwasinski, and K.J.Ray Liu,
  Pizza Party Algorithm for Distortion Management
  in Downlink Single-Cell CDMA Systems, submitted
  to Allerton Conference, 2004.
• Zhu Han, Jane Wang and K.J.Ray Liu, "A Resource
  Allocation Framework with Credit System and User
  Autonomy Over Heterogeneous Wireless Networks", 
  accepted for Globecom 2003.
• Jane Wang, Zhu Han and K.J.Ray Liu, "MIMO-OFDM
  Channel Estimation via Probabilistic Data
  Association Based TOA Estimation",  accepted for
  Globecom 2003.
• Zhu Han and K.J.Ray Liu, "Joint Power Control and
  Blind Beamforming in Wireless Networks", ICC
  2003.
• Zhu Han and K.J.Ray Liu, "Throughput Maximization
  Using Adaptive Modulation in Wireless Networks
  with Fairness Constraint ", WCNC 2003.
• Andres Kwasinski,, Zhu Han and K.J.Ray Liu,
  "Power Minimization under Real-Time Source
  Distortion Constraints in Wireless Networks",
  WCNC 2003.
• Zhu Han and K.J.Ray Liu, "Power Minimization
  Under Constant Throughput Constraint in Wireless
  Networks with Beamforming ", VTC fall, 2002.
• Zhu Han and K.J.Ray Liu, "Joint Adaptive Power
  and Modulation Management in Wireless Networks
  with Antenna Diversity", SAM 2002.
• Zhu Han and K.J.Ray Liu, " Adaptive Coding for
  Joint Power Control and Beamforming Over Wireless
  Networks", SPIE 2002.
• Zhu Han and K.J.Ray Liu, "Adaptive SINR Threshold
  Allocation for Joint Power Control and
  Beamforming over Wireless Networks", VTC fall,
  2001.