Mehmet Bilgi Department of Computer Science and Engineering 1 Capacity Scaling in Free-Space-Optical...

Mehmet Bilgi Department ofComputer Science and Engineering

Capacity Scaling in Free-Space-Optical

Mobile Ad-Hoc Networks

Mehmet Bilgi University of Nevada, Reno

2


Agenda

RF and FSO Basics FSO Propagation Model FSO in Literature Mobility Model and Alignment Simulation Results Conclusions Future Work

3


RF and FSO Illustration

Different natures of two technologies: omni-directional and directional

Omni-directional RF antenna

Directional FSO antenna

TransmitterReceiver

Transmitter

Receiver

4


A well-known fact: RF suffers from frequency saturation and RF-MANETs do not scale well √n as n is increased [1]

Linear scalability can be achieved with hierarchical cooperative MIMO [2]

imposing constraints on topology and mobility pattern Omni-directional nature of the frequency propagation causes:

Channel is a broadcast medium, overhearing Security problems Increased power consumption to reach a given range

End-to-end per-node throughput vanishes: approaches to zero as more nodes are added

1 Gupta, P. Kumar, P.R. , The capacity of wireless networks, IEEE Transactions on Information Theory, ‘00

2 Ozgur et al., Hierarchical Cooperation Achieves Optimal Capacity Scaling in Ad Hoc Networks, IEEE Transactions on Information Theory, ‘06

RF Saturation

5


Fiber Optical Solutions As of 2003;

Only ~5% of buildings have fiber connections ~75% of these buildings are within 1 mile range of fiber

Laying fiber to every house and business is costly and takes a long time Considered as sunk cost: no way to recover

Purchase land to lay fiber Digging ground

Maintenance of fiber cable is hard Modulation hardware is sensitive and expensive ISPs are uneager to deploy aggressively because of initial costs They are deploying gradually Attempts existed in near past:

California, Denver, Florida (before 2000)

1 Source: 02-146 ExParte FCC WTB Filing by Cisco Systems, May 16, 2003

6


FSO Advantages Materials: cheap LEDs or VCSELs with Photo-Detectors, commercially available,

<$1 for a transceiver pair Small (~1mm2), low weight (<1gm) Amenable to dense integration (1000+ transceivers possible in 1 sq ft) Reliable (10 years lifetime) Consume low power (100 microwatts for 10-100 Mbp) Can be modulated at high speeds (1 GHz for LEDs/VCSELs and higher for lasers) Offer highly directional beams for spatial reuse/security Propagation medium is free-space instead of fiber, no dedicated medium No license costs for bandwidth, operate at near-infrared wavelengths

7


FSO Disadvantages

FSO requires clear line-of-sight (LOS) Maintaining LOS is hard even with slight mobility Node often looses its connectivity: intermittent connectivity Loss of connectivity is different than RF’s channel fading Investigated the effects of intermittent connectivity on higher layers:

Especially TCP

8


FSO Propagation Model Atmospheric attenuation, geometric spread and obstacles contribute to BER

Atmospheric attenuation: Absorption and scattering of the laser light photons by the different aerosols and gaseous

molecules in the atmosphere

Mainly driven by fog, size of the water vapor particles are close to near-infrared wavelength

Bragg’s Law [1]:

σ is the attenuation coefficient, defined by Mie scattering:

V is the atmospheric visibility, q is the size distribution of the scattering particles whose value is dependent on the visibility

1 H. Willebrand and B. S. Ghuman. Free Space Optics. Sams Pubs, 2001. 1st Edition.

RL eA log10

q

V

550

91.3

kmVV

kmVkm

kmV

q

6,585.0

506,3.1

50,6.1

3/1

9


FSO Propagation Model Geometric spread is a function of

transmitter radius γ,

the radius of the receiver ς,

divergence angle of the transmitter θ,

the distance between the transmitting node and receiving node R [1]:

2

200log10

R

AG

Rmax (receiver radius)

Maximum range (our approximate model: “triangle + half-circle”) Maximum range

(Lambertian model)

Coverage AreaUncovered Area

R

Error in the approximate model

FSO Transmitter (e.g. LED)

FSO Receiver (e.g. PD)

Geom

etrical Spread of the B

eam

1 H. Willebrand and B. S. Ghuman. Free Space Optics. Sams Pubs, 2001. 1st Edition.

10


FSO Literature – High Speed

Terrestrial last-mile applications Roof-top deployments Metropolitan / downtown areas Point-to-point high speed links Use high-powered laser light sources Use additional beams to handle swaying of buildings Gimbals for tracking the beam Limited spatial reuse Some indoor applications with diffuse optics (more on this later)

Traditional roof-top FSO deployment

11


Free-Space-Optical Interconnects Inside the large computers to eliminate latency Short distances(1-10s cm) Remedy vibrations in the environment Use backup beams, misalignment detectors Expensive, highly-sensitive tracking instruments

Hybrid FSO/RF applications Consider FSO as a back-bone technology No one expects pure-FSO MANETs Single optical beam No effort to increase the coverage of FSO via spatial reuse

Deep space communications

1 M. Naruse et al., Real-Time Active Alignment Demonstration for Free-Space Optical Interconnections, IEEE Photonics Tech. Letters, Nov. 2001

FSO Literature – High Speed

Interconnect with misalignment detector [1]

12


Mobile FSO Communications Indoor, single room using diffuse optics Suitable for small distances Outdoor (roof-top and space) studies focus on swaying and vibration Scanning, tracking via beam steering using gimbals, mechanical auto-

tracking Instruments are slow and expensive We propose electronical steering methods

Effects of directional communication on higher layers Choudhury et al. worked on RF directionality, directional MAC Traditional flooding based routing algorithms are effected badly Directionality must be used for localization also (future work)

FSO Literature

13


Mobility Model Design an antenna with FSO transceivers to

Exploit directionality and spatial reuse Target mobility Multi-element antenna using commercially

available components

Disconnections will still occur But with a reduced amount Recoverable with special techniques

(auto-alignment circuit)

Our work: FSO in MANET context with mobility

1

2

3

4

5

6

7

8

9

10

11

12

1314

15

16

Multi-element optical

antenna design:

Honeycombed

arrays of

directional

transceivers

14


Mobility Model in NS-2

B-5 (Pos-1)

A 2

A 1

A 8

A-1

B 4

B 5

B 6

A-8

B 3

B 4

B 5

No network simulator has FSO simulation capabilities Each transceiver keeps track of its alignments

A table based implementation Alignment timers

12

3

4

7

8

65

12

3

4

7

8

65

Node-A

Node-Bin Pos-1

Node-Bin Pos-2

Node-Bin Pos-3

A-7

B 2

B 3

B 4

B-3 (Pos-3)

A 8

A 7

A 6

B-4 (Pos-2)

A 1

A 8

A 7

Alignment tables in interface 5 ofnode B and interface 1 of node A



Example scenario: 2 nodes with 8 interfaces each Node-B has relative mobility w.r.t. Node-A Observe the changes in alignment tables of 2 different transceivers in two nodes

15


Mobility Experiment

0

10

20

30

40

50

60

70

0

11

17

23

33

40

.5

51

.5

65

72

79

88

.5

97

.5

10

5

11

2

12

1

12

8

Angular Position of the Train (degree)

Lig

ht In

te

ns

ity

(lu

x)

Misaligned

Aligned

Denser packing will allow fewer interruptions (and smaller buffering), but more handoffs.

Received Light Intensity from the moving train

DetectorThreshold

Train looses and re-gains its alignment in a short amount of time: intermittent connectivity

Measured light intensity shows the connection profile

Complete disruption of the underlying physical link: different than RF fading

Auto-alignment circuitry:

Monitors the light intensity in all interfaces

Handles auto hand-off among different transceivers

Initiates the search phase

Search Phase:

When misaligned, an interfaces sends out a search signal (pre-determined bit sequence), freq of search signal

Waits for reception

When senses a search signal, responds it

Interfaces restore the data transmission phase

We want to observe TCP behaviour over FSO-MANETs

Misaligned Aligned

M UX

LOS op-am p& filter

M UX

LOSop-am p& filter

M UX

LOSop-am p& filter

M UX

LOS op-am p& filter

MUXD ata

S ink

4-To-2

Priority

Enc

ode

r

0

1

2

3

0

1

PD

LED

PD

LED

PD

LED

PD

LED

MU X

LOSop-am p& filter

MU X

LOSop-am p& filter

MU X

LOSop-am p& filter

MU X

LOSop-am p& filter

DE

MU

X

4-T

o-2

Pri

ority

Enc

oder

0

1

2

3

0

1

D ataSource

LED

PD

LED

PD

LED

PD

LED

PD

16


Simulations 49 nodes in a 7 x 7 grid Every node establishes an FTP

session to every other node: 49x48 flows 4 interfaces per node, each with its own

MAC 3000 sec simulation time Divergence angle 200 mrad Per-flow throughputs are depicted Random waypoint algorithm, conservative

mobility IEEE 802.11 MAC limitation (20 Mbps)

210 meters

210

met

ers

30 m

eter

s

30 meters

17


Stationary RF and FSO Comparison

RF and FSO comparison in stationary case, no mobility

18


Stationary RF and FSO Comparison

00.05

0.10.15

0.20.25

0.30.35

Thro

ughp

ut

(KB/

s)

Number of Interfaces

RF and FSO comparison with different number of interfaces

19


Mobile FSO: TCP is adversely affected

Mobility Effect in FSO. TCP is adversely effected.

20


Mobile RF and FSO Comparison

RF/FSO comparison w.r.t. Speed

21


Node Density Effect

Fixed power: 49 nodes Increase the separation b/w nodes and the area Keep the source transmit power same

Adjusted power: 49 nodes Increase the separation b/w nodes and the area Adjust the source transmit power so that they can

reach increased distance

22


Node Density with Fixed Power

Both performs poorly in a larger area when power is not adjusted accordingly

23


Node Density with Adjusted Power

RF performs better when power is adjusted,

Uncovered regions causes FSO’s loss

RF’s power consumption is way bigger than FSO’s

24


Mobile UDP Results

0

0.2

0.4

0.6

0.8

1

1.2

1.4

4 9

Number of Nodes

Th

rou

gh

pu

t (K

B/s

)

TCP

UDP

UDP and TCP mobile throughput comparison

25


Conclusions

FSO MANETs are possible and provides significant benefit via spatial reuse

Mobility affects TCP performance severelyRF and FSO are complementary to each

other; coverage + throughput

26


Future Work

Introduce buffers at LL and/or Network Layer Group concept Directional MAC Effect of search signal sending frequency

27


Questions

Date post:	22-Dec-2015
Category:	Documents
View:	224 times
Download:	0 times

Mehmet Bilgi Department of Computer Science and Engineering 1 Capacity Scaling in Free-Space-Optical...

Documents