Under LOS conditions

As in the first case, a scenario is assumed whereby a user is in a lightly wooded environment.

In such an environment a gradual transition can be observed with the Rice factor varying from approximately 4 dB to 12 dB [108]. In these simulations, the Rice factor is varied by 0.1 dB every 10 seconds. Even though the system is capable of operating in environments where the Rice factor varies faster than this, a slower variation is considered in order to more accurately calculate the BER (as explained in Appendix B). Fig. 7.5 shows the Eb/No requirements for the selected coding and modulation combinations to obtain a B E R = 10“ ® when the maximum Doppler frequency, fdmax is 140 Hz and the spreading factor is 32.

Table 7.3 summarises the performance of the adaptive transmission system. In this Table, C stands for coding and the following number C, e.g. 2 or 3 denotes half rate or third rate respectively. Similarly M stands for modulation with order 4 or 8 (QPSK and 8PSK). The Eb/NoReq[dB) denotes the required Eb/No in order to obtain a BER of 10“ ®. In this case a fixed power level is assumed to be transmitted which has a corresponding Eb/No at the receiver of 8.1 dB. As the Rice factor increases, the channel conditions become more favourable and hence the user bit rate can be increased without changing the power requirements. In this case, the transmitter initially uses 1/3 rate QPSK as the coding and modulation combination. When

C3M4 - 0 - C2M4 C2M8

m 10 uf 8

Rice factor (dB)

Figure 7.5: Ei^/Nq requirements for different coding/modulation combinations {B E R 10-3)

Table 7.3: PERFORMANCE OF THE SYSTEM WITH OPTIMUM PARAMETERS K

(dB)

Cod/Mod E\)lNoReq (dB)

Bit rate (kbps)

3 C3M4 8.8 64.5

5 C2M4 8.4 96

7 C2M4 6.4 96

8 C2M8 9.0 144

9 C2M8 8.3 144

11 C2M8 7.4 144

7.4. Performance evaluation of the adaptive physical layer 157

160

• Hf- Bit R ate 140

130

I

I

C3M4' 70

R ice factor (dB)

Figure 7.6: Performance of the system with a fixed E^/No value of 8.1 dB

the Rice factor becomes 5 dB the coding and modulation combination is switched to 1/2 rate QPSK and similarly at a Rice factor of 8 dB is switches to 1/2 rate 8PSK.

For a conventional system, Eb/No of 8.8 dB and 3.7 dB would be required (using the 1/3 rate QPSK scheme) in order to achieve a B E R = 10“ ^ at Rice factors 3 dB and 11 dB respectively.

These power requirements would be maintained via power control. Thus the dynamic range in Eb/No is 5.1 dB. However, in this adaptive approach, the received Eb/No requirements are kept constant which reduces the dependency on power control. In addition, in a fixed system the average throughput is 64.5 kbps whereas, using the adaptive system, this value can be increased to 110.3 kbps (see Fig. 7.6). Furthermore, when we consider our average BERs over the whole range of Rice factors, we obtain an overall average of 7.74 x 10“ “^ which is close to our target of 10“^.

However, since a constant power level is transmitted, and since the corresponding Eb/No is different to the Eb/No value required to achieve B E R = 10~^, a variability is observed in the BER. This variability in the average BERs over the range of Rice factors seen in Fig. 7.6 can be reduced by operating the system with a power control or signal quality estimation scheme which operates much less frequently than a conventional power control scheme.

A further case was simulated in order to show how to reduce the variability in BER using a slower power control scheme. The Eb/No requirements to achieve a B E R = 10“ ^ and

- f - C3M4 - e - C2M4 C2M8

■d - Power

S .

Rice factor (dB)

Figure 7.7: E\,/No requirements for B E R = 10 ^ and transmitted power levels

the power levels (in terms of Eb/Ng values at the receiver) are shown in the Fig 7.7. The maximum Doppler frequency, fdmax and the spreading factor were assumed to be 70 Hz and 32 respectively. Fig 7.8 shows the BER and user bit rates. It can be seen that the variability of the BER is much less than in the previous case. In this case, the overall average BER was found to be 7.67 x 10“^. In this case the average throughput is improved from 64.5 kbps (in a fixed system) to 106.125 kbps (in the adaptive system). The dynamic power range is 4 dB whereas it is 7.5 dB for a non adaptive system.

Another case has been simulated with a different set of parameters and the Eb /No requirements for B E R = 10“ ^ for this case is shown in the Fig. 7.9. In this case, the maximum Doppler shift and the spreading factor were assumed to be 70 Hz and 64 respectively. User bit rates corresponding to 1/3 rate QPSK, 1/2 rate QPSK and 1/2 rate SPSK are 28, 42 and 64 kbps respectively. As in the previous case, the user is assumed to be in a lightly wooded environment where the maximum rate of change of Rice factor is observed. The received power levels (in terms of Eb/No) is shown in the same figure.

Fig 7.10 shows the final result of the implemented adaptive physical layer. For this simulation, rate matching has also been included. However, puncturing is avoided and a very low repetition factor was considered in order to keep the power and spectrum efficiency high.

7.4. Performance evaluation of the adaptive physical layer 159

150

BER 140

Bit R ate

130

120

0)10 nog

100

t— • — ÿ — » -#— ■

Rice factor (dB)

Figure 7.8: BER and user bit rates in different environments

C3M4 - o - C2M4 -ar- C2M8

-o — Power

m 12 iiflO

Rice factor (dB)

Figure 7.9: Eb/No requirements for B E R = 10 ^ and transmitted power levels

bIrI:70 Bit R ate

65 60 55

S)10“

40

10'^

Rice factor (dB)

Figure 7.10: BER and user bit rates in different environments

For a conventional system E^/No of 12.8 dB and 5.1 dB vyould be required in order to achieve a B E R — 10"^ at Rice factors 3 dB and 11 dB respectively. The dynamic range of the power control is 7.7 dB. The user bit rate corresponding to a fixed system is 28 kbps whereas, using the adaptive system, this value can be increased to 46.75 kbps (see Fig. 7.6). Furthermore, when we consider our average BERs over the whole range of Rice factors, we get an overall average of 7.8 x 10“ ^ which is close to our target of 10“^. Moreover, the dynamic range of the power is 4 dB, which is approximately half as much as that of a non adaptive system.