1 802.16 work including Performance of DSP-based TX-RX emulator Contribution to WP2 and WP3 Daniele Borio, Laura Camoriano, Letizia Lo Presti
Dec 26, 2015
1
802.16 work including
Performance of DSP-based TX-RX emulator
Contribution to WP2 and WP3
Daniele Borio, Laura Camoriano, Letizia Lo Presti
2
Summary
WP2.1IEEE802.16a : a possible choice for the communication standard
WP2.2a simplified channel model
WP2.3the transmitter scheme
WP3.3Beamforming algorithms for ground antenna
The DSP
emulator
3
OFDM IEEE802.16a
Matlab simulation in a simplified scenario
The DSP experiment
Objective: Contribute to the standard choice
Analysis of the degradations due to an implementation on a real digital device
Test of the algorithm efficiency in a real time varying scenario
give indications about performances of a 802.16a based TX-RX system
WP2.1
4
Why DSP analysis?
To validate the receiver in a real-HW environment
- Real time (low rate)
- DAC and ADC effects
- DSP arithmetic
Expected results
- Feasibility study of a DSP implementation
- Analysis of the degradation of the performances due to HW limitations
WP2.1
5
OFDM suitability formobile applications
Delay spread
Fast fading
WP2.1
Handled by synchronization mechanisms
Handled by channel estimation mechanisms
Handled by OFDM large cyclic prefix
Handled by beamforming techniques
6
Other motivations
OFDM is suitable for mobile communications even if multipaths are absent
The WirelessMan group is working to an amendment of 802.16a standard for mobile communication
standard 802.16e (OFDM based)
Beamformig & ODFM for efficiently recover the LOS ray
Doppler shift adaptively compensated using correlation properties of cyclic prefix
WP2.1
7
Channel model
The channel model wasn’t completely available at the beginning of the study
Studies carried on a simplified scenario
Next steps:
WP2.2
Beamformer & ODFM performances considering a more complex channel model (MATLAB environment)
E.g. the POLITO channel model
Once the more suitable channel model will be defined -> integration in the DSP emulator
8
Considered scenario
Two paths
Doppler
OFDM RX +
AWGN
Beam formerSymbol synchronization
WP2.2
9
The Transmitter scheme
IDFT: Search for simplified algorithms (No multiplications)
Adoption of a traditional IFFT
WP2.3
exploit the QAM constellation
symmetry
10
Beamforming algorithms for ground antennas
Advantages
Negative aspects
WP3.3
Real time adaptation of the array factor-> suitability for mobile communication with a time varying channel
-> LOS ray recovery, multipaths and interfering cancellation
-> Noise reduction by spatial filtering
Cost: a smart antenna results expensive in comparison with mechanical steerable antenna
11
First two steps of the study
Analysis of the state of the art of the beamforming techniques for OFDM
Selection and design of the beamformer
Analysis of the beamformer based on Matlab and C, and tuning of the algorithm parameters.
Porting of the algorithm from Matlab to DSP
Analysis of the beam former based on Matlab and C,
and tuning of the algorithm parameters.
Porting of the algorithm from Matlab to DSP
WP3.3
12
State of the art & selection
N.A. (preamble necessary: weights in each subcarrier)
1. Medium complexity
2. Continuous mode
1. Low complexity
2. Continuous mode
1. High complexity
2. Burst mode
1. Medium complexity
2. Burst mode
1. Low complexity
2. Burst mode
N.A. N.A. N.A.
Post-FFT Pre-FFT
Post-FFTSubcarrier clustering
LMSon pilot carriers
LMSon preamble
BlindSemiblind
WP3.3
13
Third step of the study
Analysis of the state of the art of the beamforming techniques for OFDM
Selection and design of the beam former
Analysis of the beamformer based on Matlab and C, and tuning of the algorithm parameters.
Porting of the algorithm from Matlab to DSP
Analysis of the state of the art of the beam forming
techniques for OFM
Selection and design of the beamformer
Porting of the algorithm from Matlab to DSP
WP3.3
14
Simulation results
BER evaluation as a function of Eb/No.Constellation in presence of AWGN only, Eb/N0 = 10 dB
Constellation in presence of AWGN,
multi-path and Doppler shift, Eb/N0 = 10 dB
Constellation recovered by the beamformer (AWGN multi-path, Doppler shift, Eb/N0 = 10 dB)
Performances evaluation: BER curve
WP3.3
15
Other aspects
OFDM Synchronization
Doppler estimation and correction
Correction of phase distortions due to a bad clock signal recovery
WP3.3
Estimation of the Doppler shift using cyclic prefix correlation property
Correction in the frequency domain using pilot tones information
Frame synchronization based on a ML estimator
16
Fourth step of the study
Analysis of the state of the art of the beam forming techniques for OFDM
Selection and design of the beam former
Analysis of the beam former based on Matlab and C, and tuning of the algorithm parameters.
Porting of the algorithm from Matlab to DSP
Analysis of the state of the art of the beamforming
techniques for OFM
Selection and design of the beamformer
Analysis of the beamformer based on Matlab and C,
and tuning of the algorithm parameters
WP3.3
17
DSP implementation
Lab-demo
RX
DSPDAC
DSPADC
TX
Audio cable
WP3.3
18
Problems due to the audio channel
Multiplexing
Synchronization
WP3.3
How transmit several spatially diversified replicas of the same signal on a SIGLE STEREO CHANNEL
How recover the different replicas
-> exploit correlation between replicas
19
Problems due to the audio channel
Channel frequency selective
Channel “angle” selective
WP3.3
“Sample and Hold” audio
channel
Distortion dependent on
the arrival angle
20
Lab Demo WP3.3
21
Lab Demo WP3.3
22
Beamforming
Comparison
DSP
Matlab
WP3.3
23
Conclusions
The DSP implementation works
MP mitigated
Open problem: performances with a more realistic channel model
Synchronization OK Open problem: performances with a more realistic channel
model
Analysis with a planar array (to be done)
Doppler >
f