La dificultad de emprender acciones legales contra los particulares

Long term variation or slow variations of the local mean is referred to as shadowing. The mag­

nitude of the mean is described by a log-normal pdf. These shadow effects are due to trees, small buildings, and other obstructions. Due to the larger free space loss and on-board RF power scarcity, mobile satellite systems are forced to operate under LOS propagation condi­

tions, at least for medium-to-high data rates [22].

The parameters in Table 7.4 have been extracted from measurements performed in a real land mobile satellite channel [100]. These L band measurements were obtained by using IN­

MARSAT’S MSAT-A satellite. The curve 7.11 shows the envelope variation in a suburban area.

7.4. Performance evaluation o f the adaptive physical layer 161

where the streets were bordered by one- and two -storey homes and the curve 7.12 corresponds to a non forested rural area.

So far the channel considered is based on a stationary stochastic process with fixed parameters.

Lutz et el. have introduced a nonstationary model based on a two state Markov model [99]

in which the fading process is switched between a Rice process (good channel state) and a Rayleigh-lognormal process (bad channel state). They are denoted by SI and S2 (see Table 7.4). The BER curves corresponding to good states and bad states in rural and suburban areas are presented in the Fig. 7.13. It should be noted that the E^/No requirements when the user is in shadowed areas are much higher than the requirements for the non shadowed case.

Therefore, the coding and modulation combination which has the best power efficiency (i.e.

1/3 rate, QPSK) has to be used when the user is in a shadowed area. Alternative techniques such as diversity may also have to be used in order to reduce the Eb/No requirements.

When the user is in an environment as discussed above, the efficiency of the adaptive physical layer mainly depends on the time duration that the user is in the unshadowed area. As soon as the user comes out of the shadowed area, the received signal power increases significantly.

The receiver then staits estimation of the Rice factor which takes approximately 100 frames and sends it to the transmitter. Even in such circumstances, the adaptive approach is still advantageous over fixed systems due to the fact that it estimates the received signal strength and uses the optimum coding/modulation combination suited for that channel conditions.

The adaptive system proposed in this thesis is well suited for LEO systems which have a re­

duced propagation delay. However, the adaptive approach is still beneficial for MEG/GEO cases since the variation of the Rice factor is slower. This allows the system to use the se­

lected modulation and coding combination for a longer time leading to a reduced complexity by avoiding a frequent change of modulation and coding schemes.

Fig 7.14 shows user bit rate as a function of the time duration where the user is in an unshad­

owed conditions. Total time duration considered is 10 seconds. In this case, when the user is in a shadowed area, 1/3 rate QPSK (corresponding bit rate is 28 kbps) is used while 1/2 rate 8PSK (corresponding bit rate is 64 kbps) is used when the channel is considered to be a good state.

For instance, time 4 in unshadowed area means that the user transmits using 1/2 rate 8PSK for 4 seconds while he uses 1/3 rate QPSK for the remaining 6 seconds since he is in a shadowed

Table 7.4: PARAMETERS OF THE TWO-STATE MODEL FOR SUBURBAN AND RURAL

area. It should be noted that in a non adaptive type of communication system the user bit rate would always be 28 kbps irrespective of the time when user is in an unshadowed area.

The Rice factor estimation algorithm can assess the channel quality in terms of the LOS com­

ponent strength with respect to the diffused counterpart. As mentioned earlier, in the absence of the LOS component, the most power efficient coding and modulation combination has to be used (at the expense of bandwidth) in order to satisfy the Eb/No requirements. In some special cases, alternative techniques, such as diversity schemes, have to be used depending on the degree of shadowing. Although it is considered to be beyond the scope of this research work, it is interesting to evaluate the impact of the degree of shadowing in the selection of the optimum combination of modulation and coding schemes under shadowing conditions. This is a topic that needs to consider in future work.

7.5 Summary and conclusions

In this chapter, the benefits of an adaptive physical layer are investigated. The improvement of the spectrum efficiency is clearly demonstrated. In addition, the following conclusions are made from the results obtained;

• Modulation detection is performed based on the statistical characteristics of the phase distribution of the received signal. Evaluation is for the worst possible case: where modulation detection is performed for every frame and even for this case the accuracy is generally above 90%. The accuracy increases with the increase of Eb/Np as well as the

7.5. Summary and conclusions 163

-20 4 0

—60 8 0

--100 4 6

time (s)

Figure 7.11: Simulation of the envelope, r){t), for suburban channel

-20

Î

-8 0

-100

time (s)

Figure 7.12: Simulation of the envelope, for rural channel

— Rural-S1

—f— Rura!-S2 SubUrban-S1 - a - SubUrban-S2

Ui

O)

10"'

0 5 10 15 20 25 30 35

Eb/No (dB)

Figure 7.13: BER performance under shadowing conditions

E-50

45 40

35

Time in unshadowed area (s)

Figure 7.14: User bit rate as a function of channel states

7.5. Summary and conclusions 165

estimation frame window. Since the 8PSK constellation is denser than that of QPSK, the algorithm performs better with the latter.

• When the Eb/No requirements for different coding and modulation combinations have a small dynamic range, a constant power can be transmitted. This greatly reduces the power control requirements. This is a typical case for a fast moving user. The con­

stant power (or the corresponding Eb/No) value can be found by averaging the Eb/No requirements for different combinations of coding and modulation.

• For a slowly moving user, the dynamic range requirements of Eb/No increases. Vari­

ability of BER can be observed in such situations.

In adaptive systems the dynamic range of the power required is less than that of a non ! adaptive system. This is a major advantage in an adaptive system and greatly eases the

implementation compared to the non adaptive system with conventional power control.

• In order to overcome the variability of BER, a power control or signal quality estimation scheme which operates much less frequently than conventional schemes can be con­

sidered. This is still much more convenient compared to conventional power control schemes. In this case the power control bits can be replaced with the coding and modu­

lation information which are sent approximately every 100 frames.

• When the user is in a shadowed area, the system performance depends on the time dura­

tion that the user is in the unshadowed area.

• The efficiency of the adaptive system also depends on the type of constellation used. In GEO systems the variation of the Rice factor is much slower than that of a non-GEO system which means less frequent change of modulation and coding schemes.

Chapter 8