Title: QoS and Fairness Constrained Convex Optimization of Resource Allocation for Wireless Cellular and Ad
1QoS and Fairness Constrained Convex Optimization
of Resource Allocation for Wireless Cellular and
Ad Hoc Networks
- David Julian, Mung Chiang, Daniel ONelill and
Stemphen Boyd. - Jo Woon Chong
- Communication Networks Research Lab.
- Dept. of EECS, KAIST
- 2002. 12. 11.
2Contents
- Introduction
- Convex Optimization
- Wireless Cellular Networks
- Throughput
- Delay
- Wireless Ad Hoc Networks
- Throughput
- Delay
- Efficiency
- Simulation Results
- Conclusion
3Introduction
- QoS is an important and commercial issue
- Different requirements for different services
- Not satisfied with best effort transmission
- QoS in a wireless network
- A difficult problem
- Physical channel is time varying and unreliable
- A computationally efficient tool for QoS is
needed - Convex Optimization
- Related with Power control !
4Introduction
- Power control In a wireless network
- To control interference
- Indirectly control the QoS seen by users on the
network - SIR (Signal to Interference Ratio)
- Used to capture interference
- Co-channel interference
- Adjacent channel interference
- In this paper to characterize the QoS parameter
- Throughput of a particular link
5Introduction
- QoS Problems to Solve
- In Wireless Cellular Network
- Determining feasibility of a set of SIR
requirements - Maximizing SIR for a particular class of users
with lower bounds on the QoS of all other users - Satisfying queuing delay requirements for users
in various QoS class
6Introduction
- In Ad Hoc Networks
- Finding the optimum power control to maximize
overall system throughput consistent with QoS
guarantees in a fading environment. - Determining feasibility of a set of service level
agreements (SLA) under network resource
constraints. - Solving for the minimum total transmission delay
of the most time sensitive class of traffic by
optimizing over powers, capacities, and SLA terms - Maximizing the unused capacity of the network
7Introduction
- Fairness Problems to Solve
- Proportional fairness
- Minmax fairness
- A joint optimization of the fairness parameters
and the QoS criteria - Other topics
- Admission control and pricing problem
- A computationally efficient Heuristics for
optimization
8Contents
- Introduction
- Convex Optimization
- Wireless Cellular Networks
- Throughput
- Delay
- Wireless Ad Hoc Networks
- Throughput
- Delay
- Efficiency
- Simulation Results
- Conclusion
9Convex Optimization
- Optimal solution for nonlinear problems
- QoS and fairness problems are nonlinear.
- To solve efficiently
- Convex optimization
- Minimizing a convex objective function over
convex constraint sets - Ex. ) geometric program
- Primal dual interior point method
10Convex Optimization
- Geometric program
- Focus on monomial and posynomial functions
- Definitions
- Monomial is a function ,where the
domain contains all real vectors with
non-negative components - A posynomial is a sum of monomials
11Convex Optimization
- Geometric program is an optimization problem with
the following form - Minimize
- Subject to
- Where and are posynomials and are
monomials - This is not a convex optimization problem
- A change of variables is needed
- ,
12Convex Optimization
- Minimize
- Subject to
- Where are convex functions and are
affine functions - We have a convex optimization problem
13Contents
- Introduction
- Convex Optimization
- Wireless Cellular Networks
- Throughput
- Delay
- Wireless Ad Hoc Networks
- Throughput
- Delay
- Efficiency
- Simulation Results
- Conclusion
14Wireless Cellular Network
- Contents
- Problem formulations
- Interpretations of the QoS Constrained Power
Control - Proportional and Minmax Fairness Extensions
- SIR optimization simulation
- Admission control and Pricing
15Wireless Cellular Network
- Problem formulation
- Propagation model
-
- received power
- transmitted power
- reference distance for the antenna
far-field - propagation path length
- path loss exponent
- normalization constant
16Wireless Cellular Network
- SIR for link
-
- spreading factor in CDMA
- effect of normalization factor, the
effect of beamforming, and other factors
17Wireless Cellular Network
- Constraints Description
- Interference due to users, including base station
and mobiles, in index set must be smaller
than some positive constant because their
assigned QoS values are relatively low. - Interference due to users in index set has
to be smaller than the received signal power for
some mobile k so as to achieve a required SIR - The received signal power for some mobile k needs
to be exactly equal to a positive constant - As in the special case of the classical power
control scheme to solve the near-far problem in
CDMA, the received signal power for one mobile
needs to be equal to that of another mobile
18Wireless Cellular Network
- Formulations
- SIR constrained optimization of power control
- To maximize SIR for a particular user under QoS
constraints for other users in a cellular network
- A convex optimization problem
- If theres no objective function, then the above
formulation is a SIR requirement feasibility
problem
19Wireless Cellular Network
- SIR constrained optimization for minimum power
- The minimum power vector under the QoS
constraints can be determined - A weighted sum or powers, or the maximum user
power can be minimized. - SIR constrained with proportional fairness
- SIR constrained with maxmin fairness
20Wireless Cellular Network
- Admission Control and Pricing
- Spot pricing
- To charge more to users who consume more of the
total system user capacity. - Method
- The price could then be set as an linear function
of resource reduction in system - A user could experience different spot pricing
at different times depending on the existing load
on the system when the user sought to access the
network.
21Wireless Cellular Network
- Queuing delay is particularly important for
bursty digital data. - Example
- M/M/1 queue
- Link transmission rate
- Service rate
- SIR should be larger than
22Wireless Cellular Network
- M/M/1 queue
- Average queuing delay D
-
- Link transmission rate
- Service rate
- QoS agreement
- Average delay bound and average maximum arrival
rate - This bound can be met by constraining the SIR on
this link to exceed a minimum threshold, so that
link transmission rate is larger than
Wireless Cellular Network
23Contents
- Introduction
- Convex Optimization
- Wireless Cellular Networks
- Throughput
- Delay
- Wireless Ad Hoc Networks
- Throughput
- Delay
- Efficiency
- Simulation Results
- Conclusion
24Wireless Ad Hoc Network
- Contents
- Multi-hop network model and Rayleigh fading
- Outage probability and system throughput
- Throughput optimization
- Admission control and pricing
- Pricing simulation
25Wireless Ad Hoc Network
- Similar to wireless cellular network
- Except for Multi-hop
- Multi-hop network model and Rayleigh fading
- G path gain
- F fading factor
- P transmission Power
- Ex) G_ij j th transmitter to I th receiver
- Outage probability and system throughput
26Wireless Ad Hoc Network
- Outage probability
- Aggregate data rate for system
27Wireless Ad Hoc Network
- Throughput optimization
- Optimize power for throughput maximization
- Constraints Description (in order)
- The data rates demanded by existing system users
- Outage probability demanded by users using single
hop - Outage probability demanded by users using
multi-hop - regulatory or system limitation on transmitted
power
28Wireless Ad Hoc Network
- Admission Control and Pricing Pricing
Simulation - A user is admissible if a feasible solution of
the problem exists. - Consideration of multi-hop
29Wireless Ad Hoc Network
- Throughput optimization
- Weighted Joint capacity and Delay Minimization
- Constraints Description (in order)
- Link capacity constraint
- Delay guarantee constraint
- Delivery prob. constraint
- Guaranteed data rate to each class of traffic
30Wireless Ad Hoc Network
- Throughput optimization
- Weighted Joint capacity and Delay Minimization
- paramters
- K_j set of traffic using link j
- J_k set of links traversed by QoS class k
- n_k number of packets dynamically admitted in
the kth class of traffic - p_j prob. that this link will be maintained
- C_j C_j packets per sec
- b_k bandwidth
- d_k delay guarantee SLA
31Contents
- Introduction
- Convex Optimization
- Wireless Cellular Networks
- Throughput
- Delay
- Wireless Ad Hoc Networks
- Throughput
- Delay
- Efficiency
- Simulation Results
- Conclusion
32Wireless Cellular Network
- SIR optimization simulation
- Environment
- Five users in Formulation 1
- The five users are spaced at distance d of 1, 5,
10, 15, and 20 units from the base station. - Power drop off factor
-
-
- Noise power 0.5uW
- Max. power constraint 0.5W
- CDMA with 10
33Simulation Results
- Threshold SIR (x-axis)
- Optimized SIR (y-axis)
- In high threshold SIR
- In moderated threshold SIR
- In low threshold SIR
34Simulation Results
- 1st class(data)
- Path ABCD
- Rate 50 packets/s
- Max. delay 0.2
- 2nd class(data)
- Path DFEA
- Rate 50 packets/s
- Max. delay 0.2
- 3rd class(voice)
- Path ABFD
- Rate 250 packets/s
35Simulation Results
- Minimizing a weighted sum of Voice traffic delay
and the total capacity used. - Capacity increase -gt delay decrease
- Capacity decrease
- -gt delay increase, and at some point
- Saturated!!!
36Simulation Results
- Heuristic method gives similar optimal solution
and has more computational efficiency - Heuristic 1
- Heuristic 2
- Iterative method in getting outage probability
37Contents
- Introduction
- Convex Optimization
- Wireless Cellular Networks
- Throughput
- Delay
- Wireless Ad Hoc Networks
- Throughput
- Delay
- Efficiency
- Simulation Results
- Conclusion
38Conclusion
- By using convex optimization, the following
problems solved efficiently. - Various QoS provisioning problem
- Fairness problem
- Admission control and pricing problems
- 2 efficient heuristics for optimization