Download Presentation

Loading in 3 Seconds

This presentation is the property of its rightful owner.

X

Sponsored Links

- 79 Views
- Uploaded on
- Presentation posted in: General

Distributed Association Control in Shared Wireless Networks

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.

- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -

Distributed Association Control in Shared Wireless Networks

Krishna C. Garikipatiand Kang G. Shin

University of Michigan-Ann Arbor

Shared Wireless Networks

- Advantages

•Improves network coverage andcapacity

• Under-utilized APs put to use

- Modes of operation

Peer-to-peer sharing

Public sharing

Key Features

- Uncoordinated Access Points

Internet

•Ad-hoc deployment

• No global policy

ADSL

- Backhaul Limited

•Wireless capacity > wired capacity

User

- Throughput Inefficiency

•RSSI based AP selection

AP

- • Unfairness+ low bandwidth utilization

Association Control

- An important problem1

- •Control of user associations to prevent overloading and/or starvation of users

- •Crucial for the success of sharing

A

A

C

C

B

B

Throughput

Throughput

A

B

A

B

- 1“Seven Ways that HetNetsare a Cellular Paradigm Shift”, IEEE Communications Magazine, March 2013

Setup

- Variables

•Set of users,

•Set of APs,

•Association of user is

•Association vector, where

•Set of users connected to AP is

- Throughput

Backhaul capacity

•Equal for all users connected to same AP

Airtime fraction

MAC overhead

MCSRate

Association Control Problem

- Balancing throughput via user associations

• Utility Maximization

where

is defined as the proportional fair utility

•NP-hard=> intractable for large search space

- How to solve it without a central controller ?

Related Work

- Utility based approaches

- [Bejaranoet al. 03]
- Load-balancing of APs

max-min

Centralized

- [A. Kumar and V. Kumar 05]
- Optimal association of stations and APs

proportional

Centralized

- [Kauffmann et al. 07]
- Self Organization of WLANs

delay

Distributed

- [Li et al.08]
- Approx. algo. for Multi-Rate WLANs

Centralized

proportional

None of them achieve PF in a distributed way

This Work

- Feasibility of association control without global coordination

•Concept of Marginal utility

- Optimal randomized solution with probabilistic associations

•Steady state distribution:

- Sub-optimal greedyapproach with performance bounds

•Dense networks:

•Backhaul limited:

Randomized Approach

Randomized Approach

- User associates with APs probabilistically

•Connects for a random duration, scans and switches

•Generated Markov Chain:

- Desiredsteady state distribution

whereis a fixed parameter

Lemma: For every , is an increasing function in .

Moreover, as ,

Update Process

- Poisson clock

• Users have i.i.d clocks with inter-tick duration

• Scan is triggered at a clock tick

User update process

Scanning

Association

T1

T2

T3

T4

time

- Discretization

•Equivalent DTMC is where is the global

poisson clock

Update Process, e.g.,

- Gibbs sampler

•Association prob. of user at a clock tick

• One-step transition probability is

• Markov Chain is aperiodic, irreducible

• is the steady state distribution

Not distributed as user requires global information to compute

Distributed Update Process

- Objective function separation

where utility of AP is defined as

- Define Marginal Utility for each AP w.r.t user

where is set of users connected to AP except

Distributed Update Process

- New Update rule

Distributed Update Process

- New Update rule

•User can obtain locally through scanning

Current Association

Probing AP

Distributed Update Process

- New Update rule

•User can obtain locally through scanning

Current Association

Probing AP

Distributed Update Process

- New Update rule

•User makes a decision on switching

Current Association

Selects next association with prob. distribution

Distributed Update Process

- New Update rule

•User initiates reassociation with selected AP

Old Association

New Association

Completely distributedand asynchronous

Partial Information

- Marginal utility from subset of APs is known

•Due to partial scanning or probe frame losses

•Probability of knowing utility from AP is

Current Association

Probing AP

Partial Information

- Marginal utility from subset of APs is known

•Due to partial scanning or probe frame losses

•Probability of knowing utility from AP is

Theorem 1The generated Markov chain has steady

state distribution

where

Partial Information

- Marginal utility from subset of APs is known

•Due to partial scanning or probe frame losses

•Probability of knowing utility from AP is

Theorem 1The generated Markov chain has steady

state distribution

where

Theorem 2The expected utility in steady state satisfies

where and

Greedy Approach

Best Association

- User associates in a deterministic way

•Greedy approach to randomization

•At clock tick, user chooses AP

•Results in Nash Equilibriumwhich satisfies the property

for all and all

Theorem 3The Best Association converges almost surely. Every

optimal association is an equilibrium association.

Best Association

- User associates in a deterministic way

•Greedy approach to randomization

•At clock tick, user chooses AP

•Results in Nash Equilibriumwhich satisfies the property

for all and all

Theorem 3The Best Association converges almost surely. Every

optimal association is an equilibrium association.

Equilibriumstate is not easy to find

Best Association

- Two scenarios

•Users connect to same set of APs and at same PHY rate

•All APs are backhaul limited and wireless settings are irrelevant

Dense (collocated) Network

Backhaul limited

Dense Networks

- User index can be dropped

•Number of users associated with each AP,

•Utilityof AP where , are constants

Concave

Theorem 4Every equilibrium association is globally optimal,

that is

Theorem 5It takes at most N re-associations to reach equilibrium;

each user switches at most once

Backhaul limited

- Wireless parameters can be ignored

•Number of users associated with each AP,

•Each user has different neighborhood

•Utilityof AP , assume

Concave

Theorem 6Every equilibrium association satisfies the lower

bound,

Simulation

Simulation

- Performance in random topology

•Association control performs

significantly better than RSSI approach

•Partial scanning leads to slower

convergence

Greedy approach converges to almost optimal solution

Simulation

- Comparison with other distributed policies

•Slight reduction in throughput due to PF fairness

Best Association gives the highestfairness

Conclusion

- Association control in shared WLANs

•Greedy heuristic performs close to optimal

• Achievable using a distributed mechanism

- Extendable to Heterogeneous Networks ?