25
Case Study 1: - Reducing the Energy Consumption of Macro Base Station 1 Cooperative Rotation Active Cell Rotation

Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Embed Size (px)

Citation preview

Page 1: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Case Study 1: - Reducing the Energy Consumption of Macro Base Station

1

Cooperative Rotation

Active Cell Rotation

Page 2: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

2

Active Cell Rotation Scheme with Cooperation

Proposal 1: Cyclic Activation Based Dynamic Cell Configurations For Homogeneous Deployments

[2] Z. Pan, S. Shimamoto, “Cyclic Activated Power-efficient Scheme based upon Dynamic

C e l l

Configuration”, Proc. of IEEE Wireless Communications and Networking Conference

(WCNC) 2012

Page 3: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Markov Birth-Death Process• Modeling of Traffic Variance

𝑇1 =

𝑝𝑖 ∆𝑡 =𝑁

𝑖

𝜆∆𝑡 𝑖

𝑖!𝑒−𝜆∆𝑡

𝑖

1 −𝜆∆𝑡 𝑖

𝑖!𝑒−𝜆∆𝑡

𝑁−𝑖

𝑤ℎ𝑒𝑛 𝑛 → 𝑛 + 1

𝑞𝑗 ∆𝑡 =𝑁

𝑗𝑒−𝜇∆𝑡

𝑗1 − 𝑒−𝜇∆𝑡

𝑁−𝑗

𝑤ℎ𝑒𝑛 𝑛 → 𝑛 − 1

(Binomial distribution)

n: number of calls in progress N: total users per cell

𝜆: birth rate (times/hour) 𝜇: average duration (s)

• Transition Probability of 7 States

𝑇2 =

𝑃𝐼,𝐽 ∆𝑡 = 𝑖−𝑗>𝑡ℎ

𝑝𝑖 ∙ 𝑞𝑗

𝑄𝐼,𝐽 ∆𝑡 = 𝑖−𝑗<𝑡ℎ

𝑝𝑖 ∙ 𝑞𝑗

𝑅𝐼,𝐽 ∆𝑡 = 1 − 𝑃𝑖,𝑗 ∆𝑡 − 𝑄𝑖,𝑗 ∆𝑡 , 𝑖 − 𝑗 < 𝑡ℎ

𝑃𝐼,𝐽: probability of increase 𝑄𝐼,𝐽: probability of reducing

𝑅𝐼,𝐽: probability of maintain 3

Probability of birth

Probability of death

Traffic variance: 𝑣𝑖𝑗 = 𝑃𝑟 𝑖 − 𝑗

Transition Matrix of 7 States

Page 4: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Simulation Results

Power Consumption of BS Power Consumption of MT

Active Rotation: 39% savings

Cooperative Rotation: 49% savings

4

Page 5: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

5

Call Drop Rate Delay Distributions in all postponed calls

𝐸𝐶𝑅𝐵𝑆 =𝐸𝑛𝑒𝑟𝑔𝑦 𝐶𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛

𝐸𝑓𝑓𝑒𝑐𝑡𝑖𝑣𝑒 𝑆𝑦𝑠𝑡𝑒𝑚 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦=

𝑊𝑎𝑡𝑡

𝑏𝑝𝑠𝐺𝑓 =

𝑃𝑀𝑆 1 − 𝐷𝑟

𝑇ℎ𝑟𝑜𝑢𝑔ℎ𝑝𝑢𝑡=

𝐸𝐶𝑅𝑀𝑆

1 − 𝐷𝑟

Page 6: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Summary 1• Markov Modeling and Instantaneous Traffic Simulation

• Performance Evaluation

6

Poisson Arrival

Exponential Terminated

Pure Traffic Variance

(Per Cell Group)

Threshold Control

Transition Probability of 7 States

PerformanceActive

RotationCooperative

Rotation

BS PowerConsumption

39% savings 49% savings

MS Power Consumption

14dB (shadowing) 12dB

19dB (shadowing)17dB

Call Drop Rate2.5% (shadowing)

0.2%0.6% (shadowing)

0.1%

Delay 2.69 (slots) 2.01 (slots)

(W/Gbps)0.016 (shadowing)

0.0120.013 (shadowing)

0.01

(W/Mbps)1.23 (shadowing)

0.641.61 (shadowing)

1.01

Dynamic Cell Configurations

Energy-efficient Cooperation

QoS SatisfiableBS Solution

Page 7: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

MUE

MUE

FUE

FAPMUE

FAPFUE

MBS

FAP

FAP

Cross-tier interference (MBS to FUE)Cross-tier interference (FAP to MUE)

intra-tier interference (FAP to FUE)

Q. Crit ical interferences vs.Additional energy consumption

Q. Maintain QoS vs. Sudden traffic

Power adjustment of FAP Extend the coverage of FAP Exploit idle period of FAP

with adaptive sleep patterns

Case Study 2: - Homogeneous & Heterogeneous Deployments

Homogeneous Deployment

Heterogeneous Deployment

Cross-tier interference cancellation

More services by open access

7

Page 8: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Proposal 2: Cell Sizing based Energy Optimization via Sleep Mode

8

Hybrid Access for Near-indoor FUEs

Utility based power controlreduces the intra-interferencewhile sleep activation reducesthe cross-tier interference

Dynamic coverage extensionwith adaptive transmit powerof FAP in variable traffic load

Variable sleep patterns forfemto-cluster make the smallcell deployment becomesenergy-efficient.

[3] Z. Pan, S. Shimamoto, “Cell Sizing Based Energy Optimization in Joint Macro-Femto

Deployments via Sleep Activation”, Proc. of IEEE Wireless Communications and Networking

Conference (WCNC) 2013, Published

Page 9: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

CBS

MBS-2

FAP a

FAP b

MUE a

MUE b

MBS-3

MBS-1

Dynamic cell sizing

FUE a

Cross-tier interference(MBS to FUE)

Cross-tier interference (FAP to MUE)

intra-tier interference (FAP to FUE)

Sleep patterns for femto-cluster

9

Page 10: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

FAP Coverage Extension in Sleep Edge-MBSExtension without power adjustment Extension with power adjustment

10

(birth rate = 1times/h, open access rate = 4, CBS is active)

Sleep Edge MBS

Active Center MBS

CBS

Sleep Edge MBS

Active Center MBS

CBS

Power Consumption of FAP Cluster Energy Consumption Rating

Page 11: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Proposal 2 Extension:Cell Sizing based Power allocation for Sleep Two-tier Networks

11

Neuron control based poweradjustment makes thetransmit power of FAP becomemore accurate in a faster pace.

The dynamic cell sizing canadapt to the variable trafficload in a self-optimized sense.

Extended FAPs help the MBSsleep longer and make thetwo-tier system becomeenergy-efficient.

Hybrid Access for Near-indoor MUEs

[4] Z. Pan, M. Saito, J. Liu, S. Shimamoto, “Neuron Control-Based Power Adjustment

Scheme for Sleep Two-tier Cellular Networks”, IEEE Wireless Communications and

Networking Conference (WCNC), Apr. 2014

Page 12: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Neuron Control-Based Power Adjustment with Self-Optimizing

• Artificial Neuron Model– Multi-input indicator control with high speed and accuracy.

– Changing transfer function during a learning phase.

12

learning

Basic neural control process

𝑜𝑗 = φ

𝑖=0

𝑛

(𝜔𝑖𝑗𝑥𝑖 + 𝜃𝑗)

: input

: weight

𝜑 : activation function

𝑜𝑗 : output

𝑥1⋯𝑥𝑛

𝜔1𝑗 ⋯𝜔𝑛𝑗

Page 13: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Simulation Results FAP Extension in the Sleep Macrocell

Adjustment by UBPC Adjustment by NC13

Page 14: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Simulation ResultsActive Ratio of a Sleep MBS

Active Ratio vs. Power Consumption of two-tier system

Normalized Target Adjustment Ratio

𝜆: birth rate (times/h)

14

Page 15: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Summary 2

15

• Utility and Traffic Based Power AdjustmentSleep

Macrocell

FAP Coverage Extension

FAPFAP

Cluster

Efficient MBS Efficient FBS

Cross-tier Interference

Power ControlIntra-interference

HybridAccess

SleepPatterns

• Neuron Control Based Power Adjustment

𝑥1, 𝑥2⋯𝑥𝑛 : i n p u t s𝜔1, 𝜔2⋯𝜔𝑛: weights𝜃𝑖 : correction factor𝜑: sigmoid function

Proposal for FAPs Proposal for MBS

Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Coverage extension with NC and UBPC,MBS active ratio vs. two-tier system power consumption in different open access rateAdjustment target ratio vs. convergence time

• Performance Evaluations

Page 16: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Case Study 3: - Handoffs in Heterogeneous Deployments

• Desirable Handoff Features:

Maximize: Reliability, PerformanceMaintain: Seamless mobility,

Load balancingMinimize: Interference,

Number of handoffs

Handoffs in two-tier Heterogeneous Scenario

Horizontal handoff: within the same network technologyVertical handoff: within the different network technology 16

Page 17: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

RF Fingerprint Based Localization

• Radio frequency fingerprint (RFF):Identifies the device from which a radio transmission originated by looking at the properties of its transmission, including specific radio frequencies.

17Handoffs in two-tier Heterogeneous Scenario

RFF List is created by CBS

FAP update the RFF to the CBS

CBS sends the RFF list to MUEs

MUE senses and matches the received RFF with the

target FAPHandoff preparation

Page 18: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

P-Persistent Robust Handoff Decision

18

P-Persistent handoff1-Persistent handoff

𝑃𝑖𝑛 = 𝑃𝑖𝑛_1 ∙ 𝑃𝑖𝑛_2 ∙ 𝑃𝑖𝑛_3𝑃𝑖𝑛 : Handoff_in FAP probability of MUE𝑃𝑖𝑛_1 : probability of UE movement estimation𝑃𝑖𝑛_2 : probability of channel utility estimation 𝑃𝑖𝑛_3 : probability of UE velocity coordination𝑃𝑜𝑢𝑡 : Handoff_out FAP probability of MUE

𝑃𝑜𝑢𝑡 → UE movement estimation

[5] Z. Pan, M. Saito, J. Liu, S. Shimamoto, “Energy-sensitive Handoff Scheme Employing RF Fingerprint

For Cell Sizing Based Heterogeneous Cellular Networks”, Technical Report, IEICE RCS, Mar. 2014

Page 19: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

FAP Handoff-in Probability Estimation

• UE movement estimation

𝑃𝑖𝑛_1 = 𝑃 𝑥 < ∠𝐴𝑈𝐵

= 𝑐𝑑𝑓 𝑃𝑁𝑜𝑟 𝑥 = ∠𝐴𝑈𝐵 = 𝑒𝑟𝑓 ∠𝑉𝑈𝐴 − 𝑒𝑟𝑓 ∠𝑉𝑈𝐵

∠𝐴𝑈𝐵 : max angle that UE can roam into the FAP; 𝑃 𝑥 : probability of angle between the new movement direction and the last movement direction, ∠𝑥~𝑁(0,1)

• FAP channel utility estimation

(M/M/n) queuing model: 𝑃𝑖𝑛_2 = 1 −

𝜌𝑛

𝑛!∙

𝑛

𝑛−𝜌

𝑘=0𝑛−1𝜌

𝑘

𝑘!+𝜌𝑛

𝑛!∙

𝑛

𝑛−𝜌

,

𝜌 traffic intensity, 𝜌 = 𝜆 𝜇, n :maximum number of MUE severed ;

• UE velocity based coordination

𝑃𝑖𝑛_3 = 𝑠𝑖𝑔𝑚𝑜𝑖𝑑 𝑣𝑀𝐴𝑋 − 𝑡 =1

1+𝑒− 𝑣𝑀𝐴𝑋−𝑡,

𝑣𝑀𝐴𝑋 is the max velocity of UE.

∠𝑋𝑈𝑂 = ∠𝛽 + 𝜋, 𝑖𝑓∠𝛽 < 𝜋∠𝛽 − 𝜋, 𝑖𝑓∠𝛽 > 𝜋

∠𝑉𝑈𝑂 = ∠𝑋𝑈𝑂 − ∠𝑋𝑈𝑉

= ∠𝑋𝑈𝑂 − ∠𝛼

= ∠𝛽 − ∠𝛼 + 𝜋 , 𝑖𝑓∠𝛽 < 𝜋

∠𝛽 − ∠𝛼 − 𝜋 , 𝑖𝑓∠𝛽 > 𝜋

∠𝑉𝑈𝐴 = ∠𝑉𝑈𝑂 + ∠𝜑

∠𝑉𝑈𝐵 = ∠𝑉𝑈𝑂 − ∠𝜑

cos∠𝜑 =𝑈𝐴

2+ 𝑈𝑂

2− 𝑂𝐴

2

2 ∙ 𝑈𝐴 𝑈𝑂19

Page 20: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

FAP Handoff-out Probability Estimation

• UE movement estimation𝑃𝑖𝑛_1 = 𝑃 𝑥 < ∠𝐴𝑈𝐵

= 𝑐𝑑𝑓 𝑃𝑁𝑜𝑟 𝑥 = ∠𝐴𝑈𝐵 = 𝑒𝑟𝑓 ∠𝑉𝑈𝐴 − 𝑒𝑟𝑓 ∠𝑉𝑈𝐵

∠𝐴𝑈𝐵 : max angle that UE can roam out of the FAP; 𝑃 𝑥 : probability of angle between the new movement direction and the last movement direction, ∠𝑥~𝑁(0,1)

∠𝑋𝑈𝑂 = ∠𝛽 + 𝜋, 𝑖𝑓∠𝛽 < 𝜋∠𝛽 − 𝜋, 𝑖𝑓∠𝛽 > 𝜋

∠𝑉𝑈𝑂 = ∠𝑋𝑈𝑂 − ∠𝑋𝑈𝑉 = ∠𝑋𝑈𝑂 − ∠𝛼

= ∠𝛽 − ∠𝛼 + 𝜋 , 𝑖𝑓∠𝛽 < 𝜋

∠𝛽 − ∠𝛼 − 𝜋 , 𝑖𝑓∠𝛽 > 𝜋

∠𝑉𝑈𝐴 = ∠𝑉𝑈𝑂 + 𝜋 − ∠𝜑

∠𝑉𝑈𝐵 = ∠𝑉𝑈𝑂 − 𝜋 − ∠𝜑

cos 𝜋 − ∠𝜑 =𝑈𝐴

2+ 𝑈𝑂

2− 𝑂𝐴

2

2 ∙ 𝑈𝐴 𝑈𝑂

20

Page 21: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

TSince the increasing traffic

and decreasing SINR

𝜏1 𝜏2 𝜏3

FAP

FAP

MBS

UE

UE

UE

UE

UE

UE

Dynamic Sensing & Small-cell Detection

An example of dynamic sensing intervals

Sensing

𝜏𝑖: Sensing interval of 𝑀𝑈𝐸𝑖𝑘: Coefficient𝛤𝑖: SINR of 𝑀𝑈𝐸𝑖𝛤𝑇𝐻: Threshold of SINR 21

cos∠𝛼 =𝑈𝑂

2+ 𝑈𝑈′

2− 𝑈′𝑂

2

2∙ 𝑈𝑂 𝑈𝑈′, 𝑈𝑈′ = 𝑣𝜏𝑖

𝑈𝑈′ = 𝑈𝑂 ∙ cos∠𝛼 − 𝑈′𝑂2− 𝑈𝑂 ∙ cos∠𝛼

2

𝜏𝑖 = 𝑈𝑂 cos∠𝛼 − 𝑘2 − sin2 ∠𝛼

, If 𝑈′𝑂 =k 𝑈𝑂 , 𝑘 = 𝛤𝑖/𝛤𝑇𝐻

Page 22: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Pin Pout

Pin_th

Pout_th

Pin, Pout

PDF Distribution

Handoff-in Decision:

Pin >= Pin_th

Handoff-out Decision:

Pout >= Pout_th

22

Inside FAP Outside FAP

Page 23: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

23

Page 24: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Conclusion

• Scenario: Cell sizing based sleep two-tier deployments

Horizontal Handoff: FAP FAP

Vertical Handoff: MBSFAP

• Decision process: Create the RF fingerprint database (MBS)

Dynamic sensing & RF fingerprint matching (MUE)

Anticipated the available small-cell detection

P-persistent handoff

Dynamic sensing & preparation for the next handoff

• Effectiveness & Targets Less number of sensing Energy-efficient handoff decision

Less number of handoff More reliable sleep two-tier system

24

Page 25: Case Study 1 - sl.comm.waseda.ac.jp · Proposal for FAPs Proposal for MBS Coverage extension with UBPC, FAP Power consumption in different sleep patterns, Energy consumption rating

Future Considerations

• A bright future for cellular is not assured.• Rethinking “cells” in cellular is necessary.• The sustainability and reliability should be given tradeoff.• The ”optimal” way to design cellular networks is wide open for innovation. 25