3G Evolution Chapter: 5 - EIT, Electrical and Information ... Evolution Chapter:5 Wider-bandsingleband single-carrier transmission Payam Amani P A i@itlth Department of Electrical

3G EvolutionChapter: 5p 5

Wider-band single-carrierWider band single carrier transmission

Payam Amani

P A i@ it lth

Department of Electrical and Information Technology

[email protected]

3/26/2009 3G Evolution - HSPA and LTE for Mobile Broadband 11

Outline

• Why wider-band single carrier transmission?

• Equalization against radio- channel frequency selectivity

– Time domain linear equalization– Frequency domain equalization– Other equalizer strategies

• Uplink FDMA with flexible bandwidth assignment

• DFT- spread OFDM– Basic principles

DFTS OFDM i– DFTS-OFDM receiver– User multiplexing with DFTS-OFDM– Distributed DFTS-OFDM


Why wider-band single carrier transmission

• OFDM

Advantages:– Provides overal very high transmission bandwidth.– Robust to signal corruption due to radio channel frequency selectivity.g p q y y

Drawbacks:– Large variations in the instantaneous power of transmitted signal.

• Reduced power amplifier efficiency• High power amplifier cost• Critical for uplink

– Some methods to reduce this power variations discussed in chapter 4Some methods to reduce this power variations discussed in chapter 4.• Limitations on the amount of reduction in these variations.• Significant computational complexity and/or reduced link performance.

• wider-band single carrier transmission as an alternative for multicarrier transmission especially for Uplink.


Equalization against radio- channel frequency selectivityselectivity• Equalization: main method to handle signal corruption due to radio

channel frequency selectivitychannel frequency selectivity

• Time domain linear equalizationq– Rake receiver in DS-CDMA

• Channel mached filtering with filter response as the complex conjugate of the time reversed channel impulse response.p p

• Also called Maximum Ratio Combining (MRC)

)()( * ττ −= hw

• Maximizes post filter signal to noise ratio• No compensation for radio channel frequency selectivity

Transmitter

Channel ModelReceiver

)(ts)(τh

)(tn

)(tr )(τw )(ˆ ts


Time domain linear equalization

– Zero Forcing (ZF) equalizer:

– Select the receiver filter to fully compensate for the radio channel frequency selectivity.

Denotes linear convolution

1)()( =⊗ ττ wh

– Denotes linear convolution.

– Suppression of any signal corruption caused by radio channel frequency selectivity

⊗

selectivity.

– Large or potentially very large increase in noise level after equalization.

– Overal degradation of the link performance.


– Especially for channel with large variations in frequency response.

MMSE Equalization

• Time domain linear equalization

– MMSE Equalization

• A trade-off between signal corruption due to radio channel frequency selectivity and noise/ interference.

{ }2)()(ˆ tstsE −=ε

• Select a filter to minimize the mean squared error between the equalizer output and the transmitted signal.

{ })()(

– Linear equalization implemented as time discrete FIR filter.q p


Time domain linear equalization

– Complexity of such time discrete equalizer grows relatively rapidlyequalizer grows relatively rapidly with bandwidth of the sigal to be equalized:

w )(ˆ ts)(trsnT

– More wide band signal is subject t l ti l di h l

w )(

to relatively more radio channel selectivity

}

nr

– Requires the equalizer to have a larger span to be able to

1−Lw1w0w

g pcompensate it. s


– More wideband signal leads to a correspondingly higher sampling rate for the recei ed signal

snTrate for the received signal. Filter processing shall run with higher sampling rates.

w )(ˆ ts)(tr

}

nr

– High complexity in the equalization and also calculating the inverse of channel output

1−Lw1w0w

spautocorrelation matrix. ns


Frequency domain equalization

• Reduce complexity of equalization.

• Carried out block wise in block size N. snT

nr nskSkR)(tr

• Equalization as frequency domain filtering. {

• Block size N preferably selected as for some integer n to

allow for computational efficient 0wnN 2=

pradix-2 FFT/IFFT implementation of DFT/IDFT.

1−Nw0R

0s

• For channels with extensive frequency selectivity frequency domain equalization is less complex

1−NR1ˆ −Ns


complex.


• Time domain filtering implements a time discreteimplements a time discrete linear convolution.

snTnr ns

kSkR)(tr

• Frequency domain filtering corresponds to circular convolution in the time domain

{convolution in the time domain.

• First L-1 samples at the output 0w

of the frequency domain equalizer will not be identical to the corresponding output of the 1−Nw

0R0s

p g ptime domain equalizer.

1−NR1ˆ −Ns



O l f t l t L 1 l• Overlap of at least L-1 samples.

• Discard first L-1 samples atDiscard first L 1 samples at output of the frequency domain equalizer as they are also provided as the last part of the

1−≤ L

provided as the last part of the previously received / equalized block.

• Computational overhead or higher receiver complexity. CP InsertionSingle carrier signal

generationD/A

Conversion

++

)(tx

Transmitter

higher receiver complexity.

• Cyclic prefix insertion.

generation Conversion+

N Samples N+Ncp Samples

++



• Cyclic prefix channel will seem to the receiver as circular convolution over a receiver processing block of size N.

• Frequency domain taps can be calculated directly.

)(tx

• Estimate of MMSE equalizer in frequency domain.

• Overhead in power and bandwidth.

• Less overhead increase block 2

*HW K=size : channel shall be constant in duration of a block size ; upper limit for the block size

0

2 NHW

K

K +


Other equalizer strategies

D i i f db k li ti (DFE)• Decision feedback equalization (DFE)– Previously detected symbols are fed back and used to cancel the

contribution of the corresponding transmitted symbols to the overal signal corruption. Used in combination with time domain linear filtering. Also used in combination with frequency domain linear equalization.

• Minimum Likelihood (ML) Detection or Maximum Likelihood Sequence Estimator (MLSE)

Uses the entire received signal to decide on the most likely transmitted– Uses the entire received signal to decide on the most likely transmitted sequence, taking into account the impact of time dispersion on the signal.

– Viterbi Algorithm– Used widely in 2G but too complex for 3G evolution (much wider

transmission bandwidth, much more channel frequency selectivity, much higher sampling rate )


Uplink FDMA with flexible bandwidth assignmentassignment• Share uplink radio resource using plink

intra-cell multiple access scheme.cy

• High rate packet data transmissionAssign the entire system bandwidth to a terminal

Frequency

terminal.

• Burstiness of most packet data services in most cases mobile Frequency

terminals do not have anything to send in uplink. TDMA required.

• Just TDMA is not bandwidth efficient• Just TDMA is not bandwidth efficient.

• Uplink is power limited.

• Allocating the entire system bandwidth to one terminal is inefficient in terms of bandwidth utilization.


DFT Spread OFDM

• Small variations in the instantaneous power of the transmitted signal.

• Possibility of low complexity and high quality equalization in frequency domain.

• Possibility for FDMA with flexible bandwidth assignment.

• Uplink transmission scheme for LTEUplink transmission scheme for LTE.

Transmitter 0

CPOFDM (IDFT)

D/A Conversion )(txSize -M

DFT110 ,...,, −Maaa

0

0M

N


N

Basic principles

• Small variations in the instantaneous power of the transmitted signal.

• PAR distribution is independent of modulation in OFDM.• Cubic metric: a measure of the additional back off required for a certainCubic metric: a measure of the additional back off required for a certain

signal wave form relative to the back off needed for some refrence wave form.


DFTS OFDM Reciever

• DFTS-OFDM spread signal is single carrier wideband signal which will be corrupted in case of time dispersive channel.p p

• If channel is frequency selective over span of DFt, the inverse DFT at the receiver will not be able to correctly reconstruct the original block of transmitted y gsymbol.

• Need for an equalizerq

Receiver

OFDM (IDFT) Size-M

Receiver

CP removal 110 ˆ,...,ˆ,ˆ −Maaa

)(tr nrDiscard

(IDFT) Size MIDFT

removal

Discard


User multiplexing with DFTS-OFDM

• By dynamically adjusting the transmitter DFT size and the size of block of modulation symbols the nominal bandwidth of the DFTS OFDMof modulation symbols, the nominal bandwidth of the DFTS-OFDM signal can be dynamically adjusted.

• By shifting the IDFT inputs to which the DFT outputs are mapped , the exact frequency domain position of the signal to be transmitted can be adjusted.

• Allows for uplink FDMA with flexible bandwidth assignmentsAllows for uplink FDMA with flexible bandwidth assignments.

110 ˆ,...,ˆ,ˆ −Maaa)(tr

nr


User multiplexing with DFTS-OFDM

0

1M1M

0

2MM00

2M

21 MM >21 MM =


Localized and distributed DFTS-OFDM

Size-MDFT }DFT

OFDM

5=M

Localized DFTS

}

}0}0

}05=MOFDM

}0

}0}0

5M

} }0

Distributed transmission Distributed transmission

Localized Localized t i i


transmission transmission

Questions:

Thanks for your attention :

Q ti ?Questions?


Documents

3G Evolution Chapter: 5 - EIT, Electrical and Information ... Evolution Chapter:5 Wider-bandsingleband single-carrier transmission Payam Amani P A i@itlth Department of Electrical