Chapter 5 Performance Simulation
5.2 Simulation Results
16]. The multipath observation window is set as [0,108] and the presumed number of paths
is set as 4 and 8 for the two-path channel and the ITU Veh-B channel. In the tracking stage, the value of the maximum iteration number V is set as 5. In the final stage, the group size of ICI effect is set as 5 to perform ICI cancellation and data detection. The entire simulations are conducted in the equivalent baseband with an assumption that both symbol synchronization and carrier synchronization are perfect. Finally, through the simulation, the parameter
is defined as a ratio of received bit energy to the power spectral density of noise.
Wb p
b o
N
E N
5.2 Simulation Results
In the simulations, we demonstrate the performance of the proposed channel estimation methods with two methods as described in the following. Method I performs channel initialization and channel tracking, then detect data by conventional V-BLAST detection;
Method II performs channel initialization, channel tracking, and ICI estimation and
cancellation with two-dimensional V-BLAST detection.
For comparison purpose, two performance curves with ideal channel estimation are provided for reference and served as performance lower bounds in each simulation figure.
One of the performance curves is the ideal channel estimation with the assumption of that channel is quasi-static in one OFDM symbol duration, which is the simulation giving the midpoint of fading patterns of each path, and it is the performance lower bound of performing the method I, denoted as one tap equalizer (perfect CSI); the other is the ideal channel estimation with the knowledge of all fading patterns, which is the performance lower bound of performing the method II, represented as 2-D V-BLAST (perfect CSI).
Fig. 8 and Fig. 9 shows the BER performance in the two-path channel and the ITU Veh-B channel, respectively, for normalized Doppler frequency . As can be observed in Fig. 8 and Fig. 9, the performance of the method I can almost achieve the performance curve of one tap equalizer with ideal channel estimation. However, an error floor is visible when SNR is high, which is due to the ICI effect resulting from channel variations. Method II which performs ICI cancellation can eliminate the phenomenon of error floor and achieve the performance lower bound of 2-D V-BLAST (perfect CSI). In Fig. 9, there is only 1dB degradation in the required for method II at BE compared with the performance curve of the 2-D V-BLAST (perfect CSI).
*
Fd T =0.02
b/ o
E N R=10−3
Fig.10 and Fig.11 shows the BER performance in the two-path channel and the ITU
Veh-B channel, respectively, for normalized Doppler frequency . As shown in Fig. 10, at , the required for the method II is about 5dB less than that for the method I, and only 0.6dB degradation compared with the performance curve of the 2-D V-BLAST (perfect CSI). Similarly to the above results, the performance of method I can almost achieve the performance curve of one tap equalizer (perfect CSI) with slightly degradation in the required . As the SNR is high, an error floor is clearly visible. Then
with ICI cancellation, the method II significantly outperforms the method I. In Fig. 11, with ICI cancellation, the method II also provides a substantial gain in , whereas it has 2dB gap in compared with the 2-D V-BLAST (perfect CSI).
Fig. 12 shows the BER versus normalized Doppler frequency in the two-path channel at . Clearly, the BER performance gap between the method I and the method II becomes larger as normalized Doppler frequency increases up to 0.07. But as the normalized Doppler frequency increases higher, the gap between the method II and 2-D V-BLAST with perfect CSI becomes larger too. As the normalized Doppler frequency increases, the linear model for ICI estimation might not be enough to estimate the channel variations.
Fig. 13 and Fig. 14 show the BER performance in the two-path channel and the ITU Veh-B channel for normalized Doppler frequency with different group size. We can observe that with greater group size, the BER performance gets better. Fig. 13 shows that
in the two-path channel with group size = 7, the BER performance can achieve the performance curve of the 2-D V-BLAST (perfect CSI). In Fig. 14, with greater group size, the BER performance also gets better, but the improvement of BER performance becomes smaller.
We can also observe that with the ici iteration equals two, which performs ICI estimation and cancellation iteratively, the BER performance gets better too.
10-1
10 15 20 25 30 35
10-6 10-5 10-4 10-3 10-2
Eb/No(dB)
BER
Method I
one tap equalizer (perfect CSI) Method II
2-D V-BLAST (perfect CSI)
Fig. 8 BER performance in the two-path channel for normalized Doppler frequency
* =0.02 Fd T
10-1
10 15 20 25 30 35
10-6 10-5 10-4 10-3 10-2
Eb/No(dB)
BER
Method I
one tap equalizer (perfect CSI) Method II
2-D V-BLAST (perfect CSI)
Fig. 9 BER performance in the ITU Veh-B channel for normalized Doppler frequency
* =0.02 Fd T
10-1
10 15 20 25 30 35
10-5 10-4 10-3 10-2
Eb/No(dB)
BER
Method I
one tap equalizer (perfect CSI) Method II
2-D V-BLAST (perfect CSI)
Fig. 10 BER performance in the two-path channel for normalized Doppler frequency
* =0.05 Fd T
10-1
10 15 20 25 30 35
10-5 10-4 10-3 10-2
Eb/No(dB)
BER
Method I
one tap equalizer (perfect CSI) Method II
2-D V-BLAST (perfect CSI)
Fig. 11 BER performance in the ITU Veh-B channel for normalized Doppler frequency
* =0.05 Fd T
10-2
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 10-4
10-3
Fd*T
BER
Method I
one tap equalizer (perfect CSI) Method II
2-D V-BLAST (perfect CSI)
Fig. 12 BER versus normalized Doppler frequency Fd*Tin the two-path channel at / =25dB
b o
E N
10-1
10 15 20 25 30 35
10-5 10-4 10-3 10-2
Eb/No(dB)
BER
one tap equalizer (perfect CSI) 2-D V-BLAST (perfect CSI) Group size=2
Group size=3 Group size=4 Group size=5 Group size=7
Fig. 13 BER performance in the two-path channel for normalized Doppler frequency with different group size.
* =0.05 Fd T
10-1
10 15 20 25 30 35
10-5 10-4 10-3 10-2
Eb/No(dB)
BER
one tap equalizer (perfect CSI) 2-D V-BLAST (perfect CSI) Group size=3
Group size=5 Group size=7 Group size=9
Group size=9, ici iter=2
Fig. 14 BER performance in the ITU Veh-B channel for normalized Doppler frequency with different group size.
* =0.05 Fd T