Filter Bank Multitone: A Physical Layer Candidate for Cognitive - - PowerPoint PPT Presentation

Filter Bank Multitone: A Physical Layer Candidate for Cognitive Radios

• P. Amini, R. Kempter, R.R. Chen, L. Lin, B. Farhang-Boroujeny

Outline 1) Cognitive Radio → Definition → What has already been proposed? 2) Channel Sensing/Spectrum Pooling 3) Physical Layer Solutions → OFDM → Filter bank methods 4) Payload Transmission, two corner stones: → Complete central control scheme → Fully Random scheme → Current work: best of both worlds 5) Conclusion

Cognitive Radio

• Vast majority of the available spectrum has been licensed
• Given a geographical position, most of the licensed spectra

are rarely used. Motivation:

• Secondary users (SUs) may use the spectrum when the

primary users (PUs) are inactive Idea: SUs need to be intelligent: Use spectrum when it is free and give it up when PU starts transmission Consequence

What has already been proposed: Spectrum pooling Proposed systems for CR are based on on Multicarrier communication:

T.A. Weiss and F.K. Jondral, “Spectrum pooling: an innovative strategy for the enhancement

• f spectrum efficiency,” IEEE Commun. Magazine, Vol. 42, No. 3, March 2004, pp. S8 - S14.

Use Multicarrier Communications for SU to fill “holes” in the spectrum Idea

Outline 1) Cognitive Radio → Definition → What has already been proposed? 2) Channel Sensing/Spectrum Pooling 3) Physical Layer Solutions → OFDM → Filter bank methods 4) Payload Transmission, two corner stones: → Complete central control scheme → Fully Random scheme → Current work: best of both worlds 5) Conclusion

Spectrum Pooling

• Spectrum pooling mechanism needs a channel sensing

method that continuously senses the channel.

• Channel sensing should be performed with a very high

probability of correct detection (to assure very low probability

• f interference with PUs).
• Weiss et al proposed a distributed spectrum pooling protocol

where all the nodes participate in channel sensing

Spectrum Pooling/Channel Sensing Protocol

• All SUs perform detection
• The distributed allocation vector is sent to the secondary user

basestation

• The base station combines the sensing information and

sends it to all SUs

Timo Weiss, Joerg Hillenbrand, Albert Krohn, Efficient Signaling of Spectral Resources in Spectrum Pooling Systems , Proc. of the SCVT 2003, Eindhoven, Netherlands

Our system is based on this distributed spectrum pooling protocol

Outline 1) Cognitive Radio → Definition → What has already been proposed? 2) Channel Sensing/Spectrum Pooling 3) Physical Layer Solutions → OFDM → Filter bank methods 4) Payload Transmission, two corner stones: → Complete central control scheme → Fully Random scheme → Current work: best of both worlds 5) Conclusion

Methods of Multicarrier Communications: OFDM

Advantage:

• FFT at the receiver which is deployed for demodulation can also be

used for spectrum sensing Disadvantage:

• Large side lobes may lead to interference with PUs and among SUs.

Solution:

• Extension of each OFDM block with long cyclic prefix and

suffix and apply windowing.

• T. Weiss, J. Hillenbrand, A. Krohn, F. K. Jondral, “Mutual interference in OFDM-based spectrum

pooling systems,” Vehicular Technology Conference, 2004. 2004. VTC 2004-Spring. 2004

100% overhead

Result

Methods of Multicarrier Communications: Filter Multitone

Filter Multitone (FMT): originally designed for DSL channels Advantages:

• Presence of guard-bands between subcarriers allows flexible

switching between different users

• Simple to implement
• G. Cherubini, E. Eleftheriou, S. Olcer, “Filtered multitone modulation for VDSL,” in Proc. IEEE

Globecom ’99, vol. 2, pp. 1139-1144, 1999.

25% Overhead

Result

Multicarrier Communications: Cosine Modulated Multitone

Cosine modulated Multitone (CMT): applied to DSL and wireless channels Advantages:

• High bandwidth efficiency
• Blind equalization
• B. Farhang-Boroujeny, “Multicarrier modulation with blind detection capability using cosine

modulated filter banks,” IEEE Trans. Commun., vol. 51 ,no. 12, pp. 2057-2070, Dec. 2003.

No overhead

Methods of Multicarrier Communications: Channel Sensing

• In an OFDM based SU-system, FFT may be used for spectrum

sensing

• Haykin proposed the multitaper method (MTM): no leakage problem
• S. Haykin, “Cognitive radio: brain-empowered wireless communications,” IEEE Journal

Selected Areas in Communications, vol. 23, no. 3, pp. 201-220, Feb. 2005.

MTM is relatively complex

But

FFT has a leakage problem

But

We propose filter banks as an efficient tool for spectrum sensing. This is at almost no additional cost, since in our system, filter banks are running at the receiver for data transmission.

Solution:

Spectrum Sensing: Simulation Results

Dynamic range > 60 dB

Filter banks

Outline

1) Cognitive Radio

→ Definition → What has already been proposed?

2) Physical Layer Solutions

→ OFDM → Filter bank methods

3) Channel Sensing 4) Payload Transmission, two corner stones: → Complete central control scheme → Fully Random scheme → Current work: best of both worlds 5) Conclusion

Payload Transmission

• So far, no protocol has been proposed for payload

transmission in cognitive radios

• We investigate the performance of the two corner stones of

accessing schemes to gain a better understanding of the possibilities and limits of cognitive radio networks: 1) Complete central basestation control 2) Fully random accessing scheme

A practical scheme will combine methods from these two for maximum efficiency and scalability

Payload Transmission: System Model

• Channel sensing is performed every T=5ms
• SU channel access is restricted to time slots
• The total bandwidth is divided into N=512 subcarriers
• From these the number of subcarrier available to SUs is

denoted by n(t) and is known after channel sensing.

• With a bandwidth of B=400kHz for each subcarrier the total

bandwidth available to SUs is C = n(t)B

• The secondary and primary users are modeled as Poisson

processes with aggregate arrival rates λs and λp, respectively

Simulation Scenario: System Model, cont’d

• As the number of packets in the queues exceeds

P =40 packets, the SUs enter the payload transmission stage

• Packet lengths are trimodally distributed: Internet traffic
• packet sizes=50 byte, p=0.5
• packet sizes=500 byte, p=0.4
• packet sizes=1500 byte, p=0.1

Simulation Scenario: Scheduled scheme

• SU will only be accepted/can remain in the system if the

guaranteed minimum rate Rmin=400kbps over T=5ms can be sustained

• Let s denote the number of active secondary users and k is

the maximum number of active SUs in the system:

→ if k < s, s-k users are backlogged → more users can be accepted if k > s

• available bandwidth is assigned to SUs in round-robin

fashion

Ignoring signaling overhead, upper bound

• n SU data transmission

Throughput: Scheduled Scheme

Scheduling may add significant overhead

But:

All available bandwidth can be used

Pro:

Simulation Scenario: Multi-Carrier ALOHA

• Active SUs randomly select W=10 subcarriers out of free

subcarriers (multicarrier ALOHA)

• As soon as one subcarrier has been chosen by more than
• ne SU, we assume that the transmissions by all involved

SUs are lost

Most random, worst case scenario. Lower bound on performance of SU data transmission

Throughput: Random Scheme

Very Inefficient, even when PU is “off”, collisions may happen between SU

Problem:

Backlogged packets: Random Scheme

Instability issues of fully random scheme

Problem:

Comparison

No signaling

• verhead,

Very inefficient and unstable Fully Random Signaling overhead, Excluding signaling, it allows to harness channel capacity Central Control Cons Pros

We are currently working on a distributed Multi-carrier accessing scheme based on Request to Send/Clear to Send messaging, “Best of both worlds”. A realistic accessing scheme for SU payload trans- mission needs to incorporate some SU coordination

Outline

1) Cognitive Radio

→ Definition → What has already been proposed?

2) Physical Layer Solutions

→ OFDM → Filter bank methods

3) Channel Sensing 4) Payload Transmission, two corner stones: → Complete central control scheme → Fully Random scheme → Current work: best of both worlds 5) Conclusion

Conclusions: Physical Layer

• OFDM as proposed by Weiss et al. suffers from leakage

problems which leads to inefficient usage of available spectrum. Up to 100 % overhead.

• The Multitaper sensing method as proposed by Haykin does

not have leakage problem, but is complicated to implement.

• We proposed filter banks as an efficient way for spectral

analysis as well as payload transmission. Like the mutitapper, filterbanks do not suffer from leakage. No overhead

Conclusions: Payload Transmission

• a fully random accessing scheme will not deliver good system

performance: unstable

• full central base station control may add a lot of signaling
• verhead.
• Current work focuses on a multi-carrier RTS/CTS based

accessing scheme for cognitive radios

• We currently investigate in the cross-layer aspects of

Filter bank based cognitive radio systems

The End

Questions are welcome!