1. Decision Feedback Equalization in OFDM with Long Delay Spreads
Zeeshan Qureshi

2. Digital Video Broadcasting (DVB)
Rapidly becoming world-wide standard for digital TV
Being implemented in Europe, Australia, South Africa and India
Popularity due to:
Bandwidth efficiency
Improved picture/sound quality
Additional programming options
Utilizes OFDM to transmit over a Single Frequency Network (SFN) Model

3. SFN – Major Issues
Simultaneous transmission of same OFDM data from multiple transmitters
Receiver interprets a channel with a long Delay Spread
Delay spread longer than Cyclic Prefix causes increase in interference components (ISI; ICI)
Performance Degradation: Symbol recovery suffers due to interference

4. Conventional Solutions
Increase Cyclic Prefix length to Channel Delay Spread
Increases Symbol Overhead
Degrades OFDM system efficiency
Equalization in time-domain
Complex Receiver Design
Impractical to implement

5. Thesis Contribution
Implementation of a DFE-PIC based receiver configuration over OFDM channels with long delay spreads
Investigation of performance gain over conventional OFDM receiver
Evaluation done via simulations
Allows performance improvement
Maintains efficiency of OFDM
Receiver implementation is simple

6. Decision Feedback Equalizer (DFE)
DFE is a non-linear equalizer
Feedback filter: ISI cancellation using previous receiver decisions
Feed-forward filter: ICI cancellation on the transmitted symbols
Advantages:
Performance comparable to the optimum demodulator but with much lower computational complexity
Low noise enhancement

7. DFE Block Diagram _ + Feed-forward Filter Feed-back Filter
(PIC Detector)
Feed-back Filter
Decision Device
yN(t) _
SN(k)
ŜN(k)
ISI(N-1)(t)

8. Parallel Interference Canceller (PIC)
PIC detector estimates and subtracts interference for each channel in parallel
Stage-wise implementation
Stage 0 uses matched filter to estimate symbol without removal of interference
Later stages use the symbol estimates of the previous stage to estimate and remove interference components
Advantages:
Fast convergence
Low complexity

9. Interference Estimation
PIC Block Diagram
Initial Observation
Symbol Estimation
(Stages 1+)
ỹ0(t)
Ż0(k)
ŜN(k) + _
Ż0(k)
ICI(N-1)(k)
Symbol Estimation
(Stage 0 Only)
Interference Estimation
Ŝ0(k)
Ŝ(N-1)(k)

10. Simulation Environment
QPSK Modulation
OFDM Symbol Transmission
Rayleigh Channel Model
Additive White Gaussian Noise (AWGN)
Channel Delay Spread as long as the OFDM Symbol length
Perfect Channel Estimation in Receiver
Single-Tap DFE implementation

11. Implemented Scenarios
Response to System Parameter changes:
No. of ISI iterations
No. of PIC stages
Scaling of ICI components
Length of Channel Delay Spread
Performance in simulated SFN channels
Inter-site Distance between Transmitters

12. Performance Analysis Simulated SFN Channel ISD = 20 Km
Worst-case condition
0.5 % CDF

13. Final remarks Highlights of the DFE-PIC receiver
Significant performance gain achieved over OFDM receiver
Preserves OFDM system efficiency
Compensation of interference effects due to long delay spread
Simple to implement in the receiver
Low computational complexity

14. The End
Questions?
Thank You!