Channel Model

4.2 Adaptive Modulation

4.2.1 Adaptive L-PAM 4.2.2 Adaptive L-PPM 4.2.3 Adaptive M-n-PAPM 4.3 Performance under Multipath ISI

4.3.1 OOK and PAM 4.3.2 PPM and PAPM 4.4 Summary and Conclusion

4.1 Introduction

Following the discussions in Chapter 3, the desired system performance suggested the optical wireless system can benefit from employing different modulation schemes under different channel conditions. This was similar to an RF system, and a different modulation order had been used to achieve highest throughput according to SNR condition [94]. In this chapter, the performances of modulation schemes were investigated further under interference conditions. Since the trade

off between the bandwidth and power efficiency was non avoidable, modulation schemes can adaptively tune amplitude levels or pulse positions in order to maintain the maximum possible throughput under interferences [95].

As discussed in Chapter 3, the average required power to achieve a certain BER level was dependent on the power spectral density of the AWGN channel and data rate. A practical transmitter-receiver structure model can keep the transmission power constant, although the momentary signal power may vary from the average power. However, the optical wireless link can be distorted by interference from different noise source. Multipath ISI and periodic background ambient noise can contribute to the performance degradation.

Multilevel modulation schemes had the potential ability to maintain a satisfactory system performance under distortion. Rate-compatible punctured convolutional codes (RCPCs) and repetition codes (RCs) had been combined with L-PPM to give a good BER performance at the cost of lower data rates [40]. Apart from the average power requirement, the data rate and the BER were two important parameters for wireless optical links.

Wong et al analysed ISI and ambient noise impact for different modulation techniques under specific channel geometry set ups (a room size of 5m×5m×3m) [68]. Their discussions were limited to include comparison amongst OOK, 2-PPM and SIK only. In this chapter, the performance of popular modulation schemes were discussed under a more general channel model, e.g. not limited to a specific room set up, with modulation schemes extended to include L-PAM, L-PPM and

M-n-PAPM. In terms of combating the ISI and background ambient noise, the proposed adaptive modulation scheme were analysed under different channel impairments. Simulation results were used to validate the performance of the proposed scheme with other candidate for the wireless optical communication channel.

The adaptive modulation scheme here was initially intended to mitigate the data rate drop in a diffuse optical link, where multipath distortion was present [71]. However, the adaptive modulation was not limited to a diffuse model. Modulation techniques developed in this chapter were also suitable for LOS systems, where multipath distortion was not regarded as significant compared to that of diffuse systems, since the interference from ambient light noise can be reduced by increasing the optical pulse intensity, thus increasing the SNR. In the interests of data rate recovery, the optimum modulation scheme parameter under different system degradations can be obtained through searching algorisms. The candidate modulation schemes had been chosen, based on the merit of combined power and bandwidth efficiency, as detailed in Chapter 3. For model simplicity, the following assumptions were made:

a. The channel was an AWGN type

b. Synchronisation was maintained for L-PPM and M-n-PAPM

c. The system operated in an office environment (e.g. moderate radiation from the sun)

4.1.1 Channel Model

As discussed in Chapter 2, concentrating only on a specific channel can lead to loss of generality. The more general and accurate ceiling bounce model was chosen as the channel model for discussion.

The impulse response of the ceiling bounce model can be plotted versus time at a given ceiling height H, in Figure 4.1, 𝐻 = 10𝑚, time step was 1𝑛𝑠.

In Figure 4.1, the energy of the received optical pulse decreases with time, while most energy (e.g. 90%) arrived within 30𝑛𝑠 in this case, with delayed tails lasting up over to 70𝑛𝑠. The impulse response was directly related to parameter 𝑎 according to equation (2.7), and the relationships of the impulse response under different ceiling height can be obtained in Figure 4.2.

In Figure 4.2, as 𝐻 increased, the starting value of 𝑕 𝑡 decreased, which indicated that the received optical pulse energy was reduced, the energy under the 𝑕 𝑡 curve also shifted to its tail, which suggested that when optical path length increased, the delay of the pulse increased, so the ISI interferences became worse. The reverse happened when 𝐻 decreased. Thus the ceiling height can be used to reflect ISI severity.

The standard system model was derived from the OOK modulation scheme. First, for given link parameters, a corresponding normalised data rate compensation ratio was derived from the OOK scheme. Second, a multilevel modulation scheme performed a search within its available system status to find its data rate compensation ratios. Finally, comparing these ratios to the normalised OOK ratio,

the system status with best approximation to the normalised ratio represented the optimum candidate for the adaptive modulation scheme.