Equalisation Architectures for OFDM and 3G

1

• Equalisation Architectures for OFDM and 3G
• Steve McLaughlin
• Yushan Li
• David G .M. Cruickshank
• IDCOM, University of Edinburgh

2
Outline

• Introduction and Motivation
• Chip-Level Equalisation for WCDMA
• FDE in SC Systems
• Chip-Level FDE for WCDMA
• Joint structure for Channel Estimation,
  Chip-Level FDE and Parallel Interference
  Cancellation
• Conclusions

3
Motivation

• Concerned with algorithmic issues which will
  enable Multimode behaviour
• Consider UMTS and OFDM systems
• Frequency domain equalisation approach
• Unfortunately, a CP-based FDE is not compatible
  for the current UMTS system and the overhead
  introduced by a CP will reduce the spectral
  efficiency.
• Suggest some solutions to overcome this while
  minimising complexity

4
Chip-Level Equalisation for WCDMA

• RAKE The performance is dominated by the MAI and
  this results in saturation at a fairly high error
  rate.
• MUD The possibility to perform multiuser
  detection in mobile handsets is limited by its
  high complexity
• Symbol-Level Equalisation Not suitable for long
  code WCDMA.
• Chip-Level Equalisation Achieves good compromise
  between performance and complexity

5
FDE in SC Systems

• Transform the received signal from the time
  domain to the frequency domain (FFT)
• 2. Adjust each discrete frequency bins and
• make the spectrum flat. Single tap equalizer
  in the frequency domain -gt Simple structure
• 3. Transform the equalized signal back to the
  time domain (IFFT)

6
SC-FDE vs. Time Domain Equalisation Complexity

• Computationally simpler, especially for channels
  with severe delay spread (1120 chips)
• For severe channel spreading, complexity of
  frequency domain processing grows slowly than
  time domain processing.

Details may be found in Falconer, et al,
Frequency domain equalization for single-carrier
broadband wireless systems, IEEE Comm.
Magazine, April 2002
7
SC-FDE in Multimode Receivers

• Merit
• Employing a similar architecture as in OFDM
  systems, SC-FDE and OFDM can easily be configured
  to coexist, thus makes the multimode receiver
  simpler while a connection to UMTS is required

8
Issue Solution

• Issue Unfortunately, a CP-based FDE is not
  compatible for current single-carrier systems
  with no CP. It is desirable to design a receiver
  without changing the format of the transmitted
  signal.
• Solutions A number of solutions have been
  proposed for OFDM systems or single-carrier
  systems without CP. They aim at compensating the
  effect of the missing CP.
• cyclic reconstruction.

9
Solutions

• D. Kim and G. Stüber, "Residual ISI cancellation
  for OFDM with applications to HDTV broadcasting",
  IEEE Journal on Select Areas in Communications,
  Vol. 16, No. 8, Aug. 1998, pp. 1590-1599. (RISIC
  Algorithm)
• 2. C. Park and G. Im, "Efficient DMT/OFDM
  transmission with insufficient cyclic prefix",
  IEEE Communications Letters, Vol. 8, Issue 9,
  Sept. 2004, pp. 576-578.
• 3. H. Won and G. Im, "Iterative cyclic prefix
  reconstruction and channel estimation for a STBC
  OFDM system", IEEE Communications Letters, Vol.
  9, Issue 4, Apr. 2005, pp. 307-309.
• Y. Li, S. McLaughlin and D.G.M. Cruickshank,
  "Bandwidth efficient single carrier systems with
  frequency domain equalization", Electronics
  Letters, Vol. 41, No. 15, July 2005, pp. 857-858.

10
RISIC Algorithm

• Introduction In the RISIC algorithm, the
  missing CP is regarded as bursty distortion in a
  time domain block and the amount of distortion is
  diminished in an iterative process with hard
  decisions being made in the frequency domain.
• Performance degrades in channel with deep nulls
  since the hard decision will cause noise
  enhancement.

11
Existing CP reconstruction methods
RISIC Scheme and Extended RISIC Scheme
12
Chip-Level FDE for WCDMA

• In principle, proposed cyclic reconstruction
  schemes can all be extended for single-carrier
  WCDMA systems in order to deploy FDE at chip
  level. However, some of them suffer from high
  computational complexity, especially in the case
  of the application to WCDMA.
• Solutions particularly proposed for WCDMA
• Overlap-Cut Method
• FDE based on Self cyclic reconstruction
• FDE based on Slot Segmentation

13
Overlap-Cut Method

1. Applying a conventional FDE on a single carrier
   system without CP gives errors that are
   significantly larger at the edges of the block.
2. Samples at the beginning and the end of each
   equalized blocks are discarded.
3. Processing blocks are overlap with each other.

Ref M. Vollmer, M. Haardt and J. Gotze,
"Comparative study of joint detection techniques
for TD-CDMA based mobile radio systems", IEEE
Journal on Select Areas in Communications, Vol.
19, No. 8, Aug. 2001, pp. 1461-1475.
14
FDE based on Self cyclic reconstruction

• The algorithm exploits the relationship between
  the required cyclic part and the transmitted
  signal itself. The estimated cyclic part is then
  added to the received block signal to enable
  frequency domain equalization. This can be viewed
  as a cyclic reconstruction process.

Ref Y. Li, S. McLaughlin, D.G.M. Cruickshank,
"UMTS FDD frequency domain equalization based on
self cyclic reconstruction", IEEE International
Conference on Communications, Vol.3, May 2005,
pp. 2122-2126.
15
FDE based on Slot Segmentation

• By exploiting the frame and slot structures of
  the UMTS downlink, the pilots within one slot
  (for FDD mode) are used for cyclic reconstruction
  in a FDE.
• Furthermore, one slot signal is split into
  multiple segments for the sake of combating
  channel variance within one slot.

Ref Y. Li, S. McLaughlin, D.G.M. Cruickshank,
"UMTS FDD frequency domain equalization based on
slot segmentation", Proceedings of the 61st IEEE
Vehicular Technology Conference, May 2005,
Stockholm, Sweden.
16
Joint Channel Estimation, Chip-Level FDE and
Parallel Interference Cancellation structure for
WCDMA

• Accurate channel estimation for a practical
  mobile communication system is important!
• Time-multiplexed pilots require extra bandwidth
  and hence reduce bandwidth efficiency.
• Code-multiplexed pilots no bandwidth spreading
  is necessary.

17
Correlation Method

• In practice, the correlation method (CM) is a
  simple technique for channel estimation in WCDMA.
• The distorted autocorrelation property due to
  channel impairments degrades its performance.
• A high power code-multiplexed pilot sequence is
  required for better channel estimates.
• Unfortunately, high power pilot channel ? high
  MAI to the data channels.

18
IFDCE Iterative Frequency Domain Channel
Estimation

• The IFDCE method reconstructs the sum of data
  channels and the code-multiplexed pilot channel.
  The reconstructed composite signal is being
  treated as a virtual pilot signal.
• Channel estimation is performed in the frequency
  domain.
• The received WCDMA signal is equalised before
  spreading at chip level in the frequency domain.

19
Procedures

• Step 1. Correlation method is used to deliver
  initial channel estimates.
• Step 2. A RAKE receiver then operates on the
  received signal and the composite estimated
  signal is despread and hard detected.
• Step 3. K users' transmitted symbols are respread
  and rescrambled. The scrambled code-multiplexed
  pilots are added to form an estimated composite
  signal.

20
Procedures

• Step 4. The estimated composite signal and the
  initial channel estimates are used for cyclic
  reconstruction.
• Step 5. The reconstructed composite signal, being
  treated as a virtual pilot signal, is converted
  to the frequency domain and used for channel
  estimation.
• Step 6. The result from Step 5 is converted to
  the time domain and only the first L values
  (Channel is assumed to span L chips) are kept to
  form a new channel estimate.
• Step 7. Frequency domain equalisation.

21
Proposed IFDCE Structure
CR Cyclic Reconstruction
22
Parallel Interference Cancellation

• Why PIC?
• Since the proposed iterative channel estimation
  requires user symbol detection and interference
  reconstruction, a PIC is combined into the
  iterative structure in order to further enhance
  the system performance.
• The integration of PIC is with only a slight
  increase in computational complexity. This is
  exactly why the PIC is introduced into the
  iterative structure.

23
IFDCE PIC Scheme
24
Simulations

• WCDMA Systems
• Carrier Frequency 2 GHz
• Chip Rate 3.84 Mchips/s
• Spreading Factor 64
• 10 Active Users
• Pilot Channel Power 10 Whole Power
• Block Size 1024 chips (For CE and FDE)
• UMTS Vehicular A Test Channel
• Mobile Speed 30 km/h

25
Simulation Results
3.8 dB
Close to the ideal case
26
Simulation Results
Close to the single user case
27
Discussions

1. The iterative channel estimator can provide the
   PIC with better channel estimates, hence better
   performance.
2. The iterative cycle can be implemented
   efficiently in the frequency domain with
   complexity of O(NlogN) where N is the block size.

28
Conclusions

• To design a FDE for the current WCDMA system is
  very attractive. OFDM has become a strong
  candidate for the fourth generation systems and
  hence a WCDMA receiver adopting FDE will be
  compatible with the current FFT based receiver
  structures.
• By adopting FDE for single carrier WCDMA, a
  multi-mode receiver can be programmed to switch
  to a particular system more conveniently.

29

• Thank you!
• Steve McLaughlin
• IDCOM, University of Edinburgh
• Email sml_at_ee.ed.ac.uk