dynamic spectrum access (dsa) wireless networking

35
Dynamic Spectrum Access (DSA) Wireless Networking R. Chandramouli (Mouli) Thomas E. Hattrick Chair Professor Department of ECE Stevens Institute of Technology

Upload: others

Post on 04-Feb-2022

7 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Dynamic Spectrum Access (DSA) Wireless Networking

Dynamic Spectrum Access (DSA) Wireless Networking

R. Chandramouli (Mouli)

Thomas E. Hattrick Chair Professor

Department of ECE

Stevens Institute of Technology

apatel
New Stamp
Page 2: Dynamic Spectrum Access (DSA) Wireless Networking

Spectrum Regulatory Models

• Command and Control (traditional model) – Allowable spectrum use is limited by the regulatory

policy – Only licensed users are allowed to use the spectrum

• Commons Model Unlicensed secondary users can share spectrum subject to spectrum etiquettes – No guarantees on protection from interference

• Exclusive Use Model – Market-driven model – Spectrum license holder (“primary user”) can sublease

unused spectrum to a non-licensed user (“secondary user”) in time and space

– Sublease can be short term to long term – Predicted to be 70% in the near-future

2

Page 3: Dynamic Spectrum Access (DSA) Wireless Networking

Is Spectrum Scarce?

Spectrum measurement (54-88MHz) in NY City shows “white spaces” or unused spectral bands

Unused spectrum “white space”

3

Page 4: Dynamic Spectrum Access (DSA) Wireless Networking

Related Worldwide Regulatory Activities

• FCC

– Unlicensed operations in T.V. white space

• Second Report and Order and Memorandum Opinion and Order, 23 FCC Rcd 16807, Nov. 2008

• Second Memorandum Opinion and Order, FCC 10-174, Sep. 2010

• Ofcom (UK)

– T.V. white spaces • Digital dividend: Cognitive access, statement—Consulatation on

lincense-exempting cognitive devices using inter-leaved spectrum, Feb. 2009

4

Page 5: Dynamic Spectrum Access (DSA) Wireless Networking

Related Worldwide Regulatory Activities

• FCC

– ET Docket No. 10-237 (Nov. 30, 2010), NOI • Promoting more efficient use of spectrum through dynamic

Spectrum use technologies

– Incentives for dynamic spectrum use?

– Create test beds or change policies for DSA in licensed and unlicensed bands?

– Is spectrum sensing a viable technology for some bands?

– ET Docket No. 10-236 (Nov. 30, 2010), NPRM • Comments on expanding Experimental Radio Service rules to

promote research

5

Page 6: Dynamic Spectrum Access (DSA) Wireless Networking

Economics of DSA

• 85%-95% of spectrum under 3GHz is under-used (several spectrum measurement studies)

• Transition to digital TV transmission opens up prime spectrum for opportunistic use – Fewer households rely on over-the-air TV – $10 Billion/year market opportunity in TV white space

DSA+WiFi – Useful for long range wireless networks – Spectrum Bridge’s ShowMyWhiteSpace

• Low cost inter-operable first responder communications

• Co-existence among heterogeneous wireless networks (dynamic spectrum sharing/access)

• … 6

Page 7: Dynamic Spectrum Access (DSA) Wireless Networking

Use Case: Emergency Interoperable First Responder Multi-band DSA Network

WiFi 4.9GHz 3G LTE

7

Page 8: Dynamic Spectrum Access (DSA) Wireless Networking

Two Basic Ideas in DSA

• Secondary use of licensed spectrum – Primary user gets highest priority – When primary user is not using the spectrum how can

secondary detect it and opportunistically use it? – Secondary must leave the spectrum as soon as primary

user transmission begins in order to protect primary from interference

• Unlicensed spectrum (“open spectrum”) – All the users that have similar rights to spectrum – How can they detect each other’s transmission to

peacefully co-exist?

8

Page 9: Dynamic Spectrum Access (DSA) Wireless Networking

Example : Dynamic Frequency Selection in WiFi Channels 1,6,11

9

Page 10: Dynamic Spectrum Access (DSA) Wireless Networking

Protocol Stack Issues for DSA

• PHY layer

– Spectrum sensing to detect white spaces, primary user, and interference

• Detection delay (e.g., more sampling) vs. accuracy trade-off

• Detecting low SNR signals (e.g. -107dbm for wireless microphones)

– Channel bonding and fragmentation • Bond adjacent channels to obtain higher bandwidth

• Fragment a wideband channel into smaller channels

• MAC layer

– Spectrum aggregation • Aggregate non-contiguous channels for higher bandwidth

– Spectrum etiquettes • No zero-rate transmission; listen before talk, …

10

Page 11: Dynamic Spectrum Access (DSA) Wireless Networking

Protocol Stack Issues for DSA • IP layer

– Maintain IP connectivity during dynamic frequency or network switching operating in different bands

• Application layer

– Learn application traffic statistics and adapt

– Support for video streaming, VoIP etc.

– Robustness against uncertainties in spectrum availability

• Policy layer

– How to represent spectrum and usage policies?

– Policy language

11

Page 12: Dynamic Spectrum Access (DSA) Wireless Networking

Spectrum Sensing

• Sense time-varying unused spectrum – Energy detection: simple but not very reliable

– Cyclostationary detection: complex but reliable

• Requires network wide quiet periods

• Collaborative sensing – Distributed spectrum sensors detect white spaces

– Sensor decision fusion for final decision

• Wideband accurate sensing incurs delay cost

• Narrow band sensing is faster

• IEEE 802.22: coarse wideband sensing and fine narrowband sensing

• Probability of detection (90%), false alarm (10%), detection time (2s) and time to vacate channel (2s)

12

Page 13: Dynamic Spectrum Access (DSA) Wireless Networking

Soekris Engineering net5501

500 MHz AMD Geode LX CPU,

512 MB DDR-SDRAM,

4 VIA 10/100Mb Ethernet Port

2 Serial,

USB connector,

CF socket,

44 pins IDE connector,

SATA connector,

1 Mini-PCI socket,

3.3V PCI connector.

Operating System

Ubuntu 8.04

Modified open source

MadWifi drivers for

cognition enabled DSA

SpiderRadio: DSA Radio Prototype 4.9GHz public

safety band 5GHz WiFi

13

Page 14: Dynamic Spectrum Access (DSA) Wireless Networking

Exploit WNIC for Sensing? • Commercial WNICs output observed PHY errors

(comes free) – Treat WNIC as a blackbox

– PHY errors reported by WNIC when packets/signal without the intended PHY preamble is observed

• When primary user is present and transmitting – Secondary user radios present in the channel observe packets due to

different packet preamble or corrupted packet preamble (known as observed PHY errors)

– Exploit this to sense primary user transmission

• Advantage: unlike energy detection, the DSA radios need not forcefully quiet down periodically to observe/detect PHY errors

– Many practical optimization, algorithmic and implementation challenges

14

Page 15: Dynamic Spectrum Access (DSA) Wireless Networking

Normal WiFi Performance under Interference

15

Page 16: Dynamic Spectrum Access (DSA) Wireless Networking

SpiderRadio DSA WiFi under Interference

16

Page 17: Dynamic Spectrum Access (DSA) Wireless Networking

Operational Capabilities

Average Synchronization Time

Sensing 10 – 50ms, depend on precision requirement

Synchronization 4 – 18ms, depend on Network traffic congestion

Channel Switching 0.5 – 1.5ms

Channel bonding/ fragmentation

0.5 – 1ms

17

Page 18: Dynamic Spectrum Access (DSA) Wireless Networking

MAC Protocol Issues

• Multi-channel MAC : DSA radios may

operate on different channels

• Dynamic channel bonding of contiguous channels

and aggregation of non-contiguous channels

• Spectrum information distribution for MAC – Control channel based MAC

– Spectrum database based MAC

• Control channel incurs significant overhead

• Spectrum databases have to be updated constantly

• QoS guarantees very challenging

18

Page 19: Dynamic Spectrum Access (DSA) Wireless Networking

MAC Issues

• Avoid spectrum starvation (e.g., mix of broadband and narrowband users)

• Spectrum packing

• Channel bonding, fragmentation, aggregation

• Multi-MAC in a radio equipped with multiple PHY layers

• Synchronizing Tx and Rx after channel switching

• Co-existence of legacy radios and DSA radios

19

Page 20: Dynamic Spectrum Access (DSA) Wireless Networking

Channel Bonding and Fragmentation

20

Page 21: Dynamic Spectrum Access (DSA) Wireless Networking

Application Layer

• Application layer QoS may suffer

• Example: – TCP application may continue to transmit while the

physical layer tries to switch to another band

– Physical link is lost during switching band

• Erasure codes – Lost packets are erasures during channel switching

– Digital fountain codes for erasure correction

– Application layer bonding – Decide optimal channel to object mapping

– E.g., from a web page, send videos on a wideband channel and text on a narrowband channel

21

Page 22: Dynamic Spectrum Access (DSA) Wireless Networking

Medical Image Transmission

Normal WiFi under channel interference

SpiderRadio under channel interference – sense and switch

22

Page 23: Dynamic Spectrum Access (DSA) Wireless Networking

Modeling and Simulation Issues

• Lack of reliable modeling and simulation tools for DSA networks

• Few DSA network pilots and large scale field tests

• Data driven modeling and simulation of interaction from PHY to policy layer

– E.g., traffic in DSA networks : i.i.d., short-term memory, long term correlations?

23

Page 24: Dynamic Spectrum Access (DSA) Wireless Networking

Multi-radio DSA

• Devices are equipped with multiple radios • E.g., 3G and WiFi

• Current DSA technologies allow a device to connect to only one wireless network at a given time • Leads to wastage of

spectrum resources, frequent connection loss, no support for inter-operability across networks, etc.

24

Page 25: Dynamic Spectrum Access (DSA) Wireless Networking

Multi-radio DSA

• SpiderRadio prototype (multi-network aggregation)

• Enables a device to connect to multiple wireless networks simultaneously for increased reliability, data rate, security, etc.

• Uses standard WNICs

• Dynamic access to different wireless networks, different channels in a wireless network, aggregate channels across networks, etc.

• Network level sensing for DSA

• SINR

• Traffic congestion

• Security

• Cost (e.g., free WiFi vs. 3G access)

25

Page 26: Dynamic Spectrum Access (DSA) Wireless Networking

Multi-radio Multi-network Aggregation

Courtesy: Google images

One virtual aggregated broadband wireless network

LTE

WiFi

26

4.9GHz

Internet

Cloud

Page 27: Dynamic Spectrum Access (DSA) Wireless Networking

IP Layer Network Aggregation with Channel Bonding/Fragmentation

27

Page 28: Dynamic Spectrum Access (DSA) Wireless Networking

DSA Interference Mitigation in Aggregated Network

28

Page 29: Dynamic Spectrum Access (DSA) Wireless Networking

Multi-Network Aggregation Performance:

Two WiFi

2.6Mbps

without

aggregation

5Mbps

with two

WiFi network

aggregation

29

Page 30: Dynamic Spectrum Access (DSA) Wireless Networking

Security Issue Example

30

• Bonded 5.24 and

5.26 GHz channels

• Significant leakage

into other channels

• How can this be

analyzed for service

disruption attacks?

Page 31: Dynamic Spectrum Access (DSA) Wireless Networking

Two Types of Attacks

• Maximum impact attack (MAXIMP)

– Attacker tries to maximize average power leakage in each fragment

– Constraint on maximum power

– Reduces the channel capacity for the users

• Use minimum power (MINPOW)

– Attacker uses minimum power to create at least a certain level of leakage in each fragment

– Reduces the signal-to-interference-noise ratio (SINR), which, in turn, reduces throughput

31

Page 32: Dynamic Spectrum Access (DSA) Wireless Networking

Numerical Results: MAXIMP

32

IEEE 802.22 networks with N=3 Channels, K=3

fragments each, i.e., NK=9

Can reduce

capacity by

16% (almost

100 Kbps)

Page 33: Dynamic Spectrum Access (DSA) Wireless Networking

Other Challenges

• Cross-layer optimization

– How can information about the application, network and channel be used together to jointly optimize the DSA network?

• Soft-handoff capabilities

– Sensing based dynamic load balancing between the multiple bonded/aggregated wireless networks

• Underlay transmission for dynamic spectrum access

• Channel fragmentation to minimize need to move to other channels within a network

– Minimizes delay cost due to channel hopping

33

Page 34: Dynamic Spectrum Access (DSA) Wireless Networking

Research Challenges

• DSA aggregation of 3G/4G LTE, 4.9GHz, 900MHz and WiFi

• Support for simultaneous VoIP and video streaming over DSA networks

• Security features such as VPN (virtual private network)

• Support for robust DSA connectivity in mobile networks

• Low cost platform

34