POWER CONTROL ASSISTED ADAPTIVE MODULATION 87 PCZ - Threshold-based Power Control Zone 11 - 14 Adaptive Modulation Switching Thresholds K - Maximum dynamic range of Threshold-based powe
Trang 1Adaptive Modulation
In this chapter, the concept of Adaptive Quadrature Amplitude Modulation (AQAM) is intro- duced, whereby the modulation mode is adapted at the transmitter on a burst by burst basis This adaptation is implemented based on the receiver’s perceived channel quality and its main motivation is to maximise the transmission throughput at a given target BER
In our investigations AQAM is initially applied in a narrow-band environment in con- junction with power control, where the transmitted power is only varied near the modem mode switching thresholds of the AQAM scheme Subsequently, AQAM is investigated in a wideband channel in conjunction with a DE-assisted receiver In this context, a new chan- nel quality metric is proposed in order to control the choice of AQAM modes Let us now commence with a brief overview of the AQAM scheme
4.1 Adaptive Modulation for Narrow-Band
Fading Channels
A brief overview of the principles of AQAM in a narrow-band Rayleigh fading channel en- vironment is given here In a narrow-band channel, as a result of its rapid fading, the short term SNR can be severely degraded, especially if the channel exhibits a deep fade The gen- eral philosophy of AQAM is to employ a higher-order modulation mode, when the channel quality is favourable in order to increase the transmission throughput and conversely, a more robust lower-order modulation mode is invoked, when the channel quality is low This is
achieved at a constant symbol-rate, regardless of the modulation mode selected and hence
at a constant bandwidth requirement Therefore the impact of AQAM mode switching on the system’s design remains as low as possible
The concept of invoking AQAM is hence to a certain extent analogous to employing power control schemes, which are typically used to combat the effects of pathloss and slow fading However, whilst power up in order to compensate for degrading channel conditions may inflict increased cochannel interference upon other users, which in turn may require fur- ther power increments for maintaining the target quality, AQAM accommodates these chan- nel quality fluctuations without disadvantaging other users in the system AQAM can also
81
Adaptive Wireless Tranceivers
L Hanzo, C.H Wong, M.S Yee Copyright © 2002 John Wiley & Sons Ltd ISBNs: 0-470-84689-5 (Hardback); 0-470-84776-X (Electronic)
Trang 282 CHAPTER 4 ADAPTIVE MODULATION
be applied in order to support users, when traversing a variety of propagation environments, such as indoor or outdoor environments In a friendly propagation environment, where the impact of IS1 and co-channel interference is significantly lower, than in an outdoor environ- ment [146], the less robust higher-order modulation modes can be activated more frequently This results in a higher average transmission throughput, while ensuring an acceptable BER performance By contrast, the more robust lower-order modulation modes can be utilized more frequently in a more hostile outdoor environment These characteristics of the AQAM switching regime will be shown more explicitly in Section 4.3.4
In adapting the modulation mode, a signalling regime has to be implemented, in order
to harmonise the operation of the transmitter and receiver with regards to the adaptive mo- dem mode parameters In this respect, the Time Division Duplex (TDD) scheme [4,146] is employed in order to implement an open loop channel quality signalling system Unlike in Frequency Division Duplex (FDD), where the uplink (UL) and downlink (DL) transmission frequency bands are different, the UL and DL transmissions of the TDD scheme are time multiplexed onto the same carrier This results in a correlation in the fading characteristics
of the UL and DL propagation channels, yielding near-reciprocal channel conditions [ 1461, provided that the channel quality is slowly varying In a loose sense, we will often use the term reciprocity in order to indicate the similarity of the UL and DL in TDD environments Consequently an open loop signalling system can be implemented, where the modulation mode can be adapted at the transceiver based on the information acquired during its receiving mode This open loop system is encapsulated in Figure 4.1 (a) In contrast, if near-reciprocal channel conditions are not applicable, a closed-loop based signalling system shown in Figure 4.l(b) can be implemented in a FDD based system These signalling regimes will be further elaborated in Section 7.2
Having discussed briefly the principle of AQAM and the possible scenarios where it can
be applied, this leads us to explore the criterion and methodology of selecting the transmit- ter’s modulation mode The criterion used by Torrance [ 1451 was the instantaneous received power, which was estimated by exploiting the reciprocal nature of the channel in a TDD envi- ronment This estimate was then used to select a suitable modulation mode by comparing the channel quality estimate against a set of switching threshold levels l,, as depicted in Figure 4.2 For example, if the estimated instantaneous received power was between the values of I 1
and 12, according to Figure 4.2, BPSK was chosen for the next transmission burst However, when the received near-instantaneous power was below 11, where the channel was in a deep fade, the transmission was disabled This was termed as the transmission blocking mode AQAM is not only used to combat the fading effects of a narrow-band channel, but it also attempts to maximize the transmission throughput This is achieved, when a higher order modulation mode is used, if the short term SNR is favourable Conversely, the scheme also attempts to optimize the mean BER by employing a more robust modulation mode, when the channel quality is degraded As a result, there is a trade-off between the mean BER and Bits per Symbol (BPS) performance This trade-off is governed by the values of the switching
thresholds 1, As the values decrease, the probability of employing higher-order modulation
modes increases, thus yielding a better BPS performance Conversely, if the values of 1,
are increased, lower-order modulation modes are employed more frequently, resulting in an improved mean BER performance In the next section, we will review some of the advances achieved using AQAM
Trang 34.1 ADAPTIVE MODULATION FOR NARROW-BAND FADING CHANNELS 83
of local Tx
Signal modem mode used by Tx
(a) Open-loop based signalling
(b) Closed-loop based signalling
Figure 4.1: Closed- and open-loop signalling regimes for the AQAM schemes, where BS represents
the Base Station, MS denotes the Mobile Station, the transmitter is represented by Tx and the receiver is denoted by Rx
Short Term Received SNR
Figure 4.2: Stylised profile of the short term received SNR in a narrow-band channel, which is used to
choose the next modulation mode
Trang 484 CHAPTER 4 ADAPTIVE MODULATION
4.1.1 Literature Review on Adaptive Modulation
Since we have discussed the basic principles of AQAM, we can now describe the advances achieved so far in this field Historically, interest in techniques of adapting the modulation and transmission rate parameters began in 1968, when Hayes [ 1471 adapted the signal amplitude according to the prevalent channel environment by utilizing a feedback channel between the transmitter and receiver that was assumed noiseless and free from latency The adaptation of the transmission rate was then recorded by Cavers [9] in a slow Rayleigh fading environment, where the adaptation parameters were transported, again via a feedback channel
Recent work was pioneered by Webb and Steele [ l ] , where the modulation adaptation was analysed in a flat Rayleigh fading environment with applications in a Digital European Cord- less Telecommunications (DECT)-like system Star QAM [4] was used instead of Square QAM and the channel’s reciprocity was exploited in a TDD scenario in order to adapt the modulation parameters The metric used to quantify the channel quality was the received sig- nal strength and the BER Reference [ l ] also recorded the effects of block size, fading rate and
co-channel interference on the performance Slightly later, Sampei, Komaki and Morinaga introduced a variable-rate, variable-modulation-mode adaptive modulation scheme [ 1481 In this paper, an extended symbol duration of 2 rate QPSK, $ rate QPSK, full-rate QPSK, 16QAM and 64QAM were used for adaptation in a narrow-band channel environment The modulation modes and rate were switched according to the signal to co-channel interfer- ence ratio and the expected delay-spread of the channel The signal to co-channel interfer- ence ratio was estimated perfectly and the modulation control parameters were accessed via the control channels in a TDMA scenario The channel assumed a slowly varying statistic and the normalized delay spread was less than unity The results were recorded in terms
of spectral efficiency and BER performance for different cellular configurations and it was concluded that adaptive modulation showed promising advantages, when compared to fixed- mode modulation in terms of spectral efficiency, BER performance and robustness against channel delay-spread In another contribution, the numerical upper bound performance of adaptive modulation in a slow Rayleigh fading channel was then evaluated by Torrance et
al [31] and subsequently, the optimisation of the switching threshold levels using Powell minimization [ 1431 was proposed, in order to achieve a certain target performance [26] Subsequent papers were published with more emphasis on the system aspects of adaptive modulation in a narrow-band environment A reliable method of transmitting the modulation
control parameters was proposed by Otsuki et al [21], where the parameters were embedded
in the transmission frame’s mid-amble using Walsh codes Subsequently, at the receiver the Walsh sequences were decoded using maximum likelihood detection Another technique of
estimating the required modulation mode used was proposed by Torrance et al [37], where the modulation control symbols were represented by unequal error protection 5-PSK sym- bols The adaptive modulation philosophy was then extended to the wideband multi-path environments by Kamio et al [39] by utilizing a bi-directional DFE in a micro- and macro- cellular environment This equalization technique employed both forward and backward ori- ented channel estimation based on the pre-amble and post-amble symbols in the transmitted frame Equalizer tap gain interpolation across the transmitted frame was also utilized in order
to reduce the complexity in conjunction with space diversity [39] The authors concluded that the cell radius could be enlarged in a macro-cellular system and a higher area-spectral effi- ciency could be attained for micro-cellular environments by utilizing adaptive modulation
Trang 54.1 ADAPTIVE MODULATION FOR NARROW-BAND FADING CHANNELS 85
The latency effect, which occurred when the input data rate was higher than the instanta- neous transmission throughput, was studied and solutions were formulated using frequency hopping [40] and statistical multiplexing, where the number of slots allocated to a user was adaptively controlled [41]
In reference [42] symbol rate adaptive modulation was applied, where the symbol rate
or the number of modulation levels was adapted by using +-rate 16QAM, :-rate 16QAM,
;-rate 16QAM as well as full-rate 16QAM and the criterion used to adapt the modem modes was based on the instantaneous received signal to noise ratio and channel delay spread The slowly varying channel was rendered near-reciprocal by utilizing short frame duration TDD and the maximum normalized delay spread simulated was 0.1 A variable channel coding
rate was then introduced by Matsuoka et al in conjunction with adaptive modulation in Ref-
erence [34], where the transmitted burst incorporated an outer Reed Solomon code and an inner convolutional code in order to achieve a higher quality data transmission The cod- ing rate was varied according to the prevalent channel quality using the same method, as in adaptive modulation in order to achieve a certain target BER performance A so-called chan- nel margin was introduced in this paper, which basically changed the switching thresholds
in order to incorporate the effects of channel quality estimation errors The utilization of channel coding in conjunction with adaptive modulation in a narrow-band environment was also recorded by Chua and Goldsmith [35] In this contribution, trellis and lattice codes were used without channel interleaving, invoking a feedback path between the transmitter and re- ceiver for modem mode control purposes The effects of the delay in the feedback path on the adaptive modem’s performance were studied and this scheme exhibited a higher spectral efficiency, when compared to the non-adaptive trellis coded performance
Subsequent contributions incorporated space-diversity and power-adaptation in conjunc- tion with adaptive modulation, for example by Suzuki et al in order to combat effects of
the multi-path channel environment in Reference [54] at a lOMbits/s transmission rate The maximum delay-spread used was one symbol duration for a target mean BER performance
of 0.1% This was achieved in a TDMA scenario, where the channel estimates were pre- dicted based on the extrapolation of previous channel quality estimates Variable transmitted power was then applied in combination with adaptive modulation in Reference [36], where the transmission rate and power adaptation was optimized in order to achieve an increased spectral efficiency In this treatise, a slowly varying channel was assumed and the instanta- neous received power required in order to achieve a certain upper bound performance was known prior to transmission Power control in conjunction with a pre-distortion type non- linear power amplifier compensator was studied in the context of adaptive modulation in
Reference [55] This method was used to mitigate the non-linearity effects associated with the power amplifier, when QAM modulators were used
Results were also recorded concerning the performance of adaptive modulation in dif- ferent multiple access schemes in a narrow-band channel environment In a TDMA system, dynamic channel assignment was employed by Ikeda et al., where in addition to assigning
a different modulation mode to a different channel quality, priority was always given to the users in obtaining the time-slots, which benefitted from the best channel quality [56] The performance was compared to fixed channel assignment systems, where gains were achieved
in terms of system capacity Furthermore, a lower call termination probability was recorded However, the probability of intra-cell hand-off increased as a result of the dynamic channel assignment scheme, which constantly searched for a high-quality, high-throughput time-slot
Trang 686 CHAPTER 4 ADAPTIVE MODULATION
for the existing active users The application of adaptive modulation in packet transmission
was introduced by Ue, Sampei and Morinaga [57], where the results showed improved data
throughput Recently, the performance of adaptive modulation in an automatic repeat request (ARQ) system was published in Reference [ 5 8 ] , where the transmitted bits were encoded
using a cyclic redundant code (CRC) and a convolutional punctured code in order to increase the data throughput
A recent treatise was published by Sampei, Morinaga and Hamaguchi [59] on laboratory
test results concerning the utilization of adaptive modulation in a TDD scenario, where the
modem mode switching criterion was based on the signal to noise ratio and on the normalized delay-spread In these experimental results, the channel quality estimation errors degraded
the performance and consequently a channel estimation error margin was devised, in order
to mitigate this degradation Explicitly, the channel estimation error margin was defined as the measure of how much extra protection margin must be added to the switching threshold levels, in order to minimise the effects of the channel estimation error The delay-spread also degraded the performance due to the associated irreducible BER, which was not compensated
by the receiver However, the performance of the adaptive scheme in a delay-spread channel environment was better than that of fixed modulation scheme Lastly, the experiment also
concluded that the AQAM scheme can be operated for f d = lOHz with a normalized delay
spread of 0.1 or for f d = 14Hz with a normalized delay spread of 0.02, which produced a mean BER of 0.1% at a transmission rate of 1 Mbits/s In this respect, f d was the Doppler
frequency
With the above background on adaptive modulation, we shall now investigate the applica- tion of power control near the switching threshold levels, which i s termed here as threshold- based power control
4.2 Power Control Assisted Adaptive Modulation Over
Narrow-band Rayleigh Fading Channels
In this section power control is utilized in conjunction with AQAM over a narrow-band
Rayleigh fading channel, where its benefits as well as disadvantages are analysed In this
discourse, perfect power control is assumed Furthermore, we will show that a maximum
power control range of *2 dB around the modem mode switching thresholds is sufficient
in our analysis of the effects of threshold-based power control on AQAM, an issue to be
clarified during our further discourse Threshold-based power control is only applied, when
the expected received power i s within a certain range of the AQAM switching thresholds
This is best explained graphically by referring to Figure 4.3 Power control is only applied,
when the expected received SNR is within a certain range of the AQAM switching thresholds
11 - 14, and this range is denoted by the Power Control Zone (PCZ) in Figure 4.3 The width
of this range is controlled by the power control’s maximum dynamic range LC, which is also depicted in Figure 4.3 Thus, if the expected received SNR is within the PCZ of Figure 4.3,
power control is applied, where the transmitted power can be increased or decreased within
the maximum dynamic range LC or alternatively, the transmitted power can be left unchanged
The main purpose of employing the threshold-based power control scheme is to optimize the system performance of AQAM, where for example, if the expected received SNR level is
Trang 74.2 POWER CONTROL ASSISTED ADAPTIVE MODULATION 87
PCZ - Threshold-based Power Control Zone
11 - 14 Adaptive Modulation Switching Thresholds
K - Maximum dynamic range of Threshold-based power control
Figure 4.3: Schematic of the threshold-based power control scheme depicting the power control zone
(PCZ), where power control is applied The power control zones are defined by the switch- ing thresholds,l,, and the maximum dynamic range of the threshold-based power control scheme n
just below a particular adaptive switching threshold, the transmitted power can be increased
to ensure that the actual received SNR level is above that particular adaptive threshold level Consequently, a higher-order modulation mode can be used, hence increasing the throughput
of the system Alternatively, provided that the expected received SNR level is just above the adaptive switching threshold, the transmitted power can be decreased sufficiently, in order to utilize a lower-order modulation mode, thus ensuring an improved BER performance An-
other possible benefit of employing the threshold-based power control scheme is to reduce the modulation mode switching frequency of the transmitter This scheme can be utilized in order to maintain the previous modulation mode by increasing or decreasing the transmitted power, whenever the expected received SNR switching level is within the power control zone
In summary, the proposed threshold-based power control can be utilized in order to improve the AQAM performance in terms of its mean BER, mean BPS and modulation switching
Trang 888 CHAPTER 4 ADAPTIVE MODULATION
and this is used as a measure of the added complexity needed to implement the threshold-
based power control scheme Similarly, the modulation mode switching frequency is defined
as the average relative frequency of the transmitter changing its modulation mode with respect
to its previous modulation mode This provided a measure of the frequency at which the
modulation mode of the transmitter is changed In the following experiments, the threshold- based power control scheme is investigated in three different scenarios in order to achieve
firstly, an improved mean BER performance and secondly, a better BPS throughput Finally, the scheme is also invoked in order to achieve a lower reduced modulation mode switching
frequency
These experiments were conducted with varying maximum dynamic power control range
of i 0 5 d B , *l.OdB and *1.5dB Since we were mainly interested in the effects of the power control scheme, the adaptive modem mode switching thresholds were not optimized and were assigned as follows : 11 = - 0 0 , 1 2 = 8dB, l3 = 14dB, and 14 = 20dB for the non-blocking
AQAM scheme, where data was constantly transmitted Similarly, for the blocking AQAM
scheme the adaptive threshold levels were l 1 = 5dB, 12 = 8dB, 13 = 14dB, and 14 = 20dB, where the transmitter was disabled, when the instantaneous power was below l I Perfect
narrow-band channel quality estimation and compensation was assumed at the receiver Let
us now concentrate on our experimental results
4.2.1 Threshold-based Power Control Designed for an
Improved Bit Error Ratio Performance
In this section the proposed threshold-based power control scheme was optimized in order to achieve an improved mean BER performance As a result of this design criterion, whenever the expected received SNR level was above the switching threshold, but within the power
control’s dynamic range on Figure 4.3, the transmitted power was reduced in order to ensure
that a lower-order modulation mode was employed As a result, the BER was improved due
to the employment of a more robust modulation mode
On the basis of this criterion, an AQAM mode transition table can be formulated, as seen
in Table 4.1, which can be studied with reference to Figure 4.3 The adaptive switching
thresholds and the PCZ, where the power control scheme can be employed is specified in the transition table The width of the PCZ depends on the maximum dynamic range, where the
higher the range, the wider the PCZ The allocation of the present chosen modulation mode was based on the expected received instantaneous SNR in conjunction with the threshold-
based power control scheme as well as on the previous modulation mode Again, the power control mechanism was characterized in Table 4 I by the arrow notations T and 1, indicating the powering-up mode and powering-down mode, respectively
The mean BER and BPS performance of our threshold-based power control assisted
AQAM scheme for both non-blocking and transmission blocking scenarios is depicted in Figures 4.4(a) and 4.4(b), respectively, where its performance was quantified for different dynamic ranges K The performance was also compared to that of the conventional AQAM
scheme without power control As expected the BER performance of the AQAM scheme with power control improved, when compared to the conventional AQAM scheme, although the
mean BPS performance was slightly degraded This characteristic was observed for both the transmission-blocking and the non-blocking scenarios, which manifested another example of the trade-off between the mean BER and BPS performance The mean BER performance of
Trang 94.2 POWER CONTROL ASSISTED ADAPTIVE MODULATION 89
Table 4.1: The AQAM transition table, which was designed in order to achieve a low mean BER The
previous modulation mode and the expected instantaneous SNR was stated and used in order
to select the present modulation mode The notations T and .j, indicated the powering-up and powering-down mode, respectively and K represented the maximum dynamic range of the threshold-based power control scheme
the blocking scheme shown in Figure 4.4(b) was lower, than that of the non-blocking scheme for channel SNRs less than 20dB This was due to the employment of the no transmission mode (NO TX) in the blocking scheme, where the transmission was disabled until the chan- nel quality became more favourable The other notable characteristic was that as the power control’s dynamic range was increased, the mean BER performance improved This was
consistent with our expectations, since the power control zone was wider as a result of the increasing dynamic range Consequently, the threshold-based power control scheme could
be applied over a wider range of instantaneous SNRs, thus facilitating the employment of a more robust modulation mode, which resulted in a reduced BER
The modulation mode switching relative frequency was also analysed and the associated results are shown in Figure 4 3 a ) The relative switching frequency was approximately the same as that of the conventional AQAM scheme, since the threshold-based power control assisted AQAM scheme was not optimized in order to influence the switching frequency, i.e the present modulation mode was chosen on the basis of the channel quality, irrespective of the previous modulation mode A high switching probability was observed at an average channel SNR of between lOdB and 25dB, as evidenced by Figure 4.5(a) This corresponded
to the channel SNR range, where most of the switching was expected to occur Furthermore, the dynamic range K , had only a slight influence on the switching performance, as depicted
in Figure 4.5(a) When studying our modulation mode switching relative frequency results,
it is worthwhile noting that a 100% switching utilization corresponds to the event that the modulation mode is switched for every new transmission burst However, the power control utilization frequency increased, as the dynamic range increased This was attributed to the
Trang 1090 CHAPTER 4 ADAPTIVE MODULATION
wider power control zone, where the transmitted power can be increased or decreased more frequently This is evidenced by the results shown in Figure 4.5(b)
In summary, the employment of threshold-based power control in order to improve the BER performance resulted in a slight degradation of the BPS performance, while maintaining
a near constant switching frequency In the next section the scheme was optimized in order
to achieve an improved BPS performance
l * Mean BER - no power control I 0 Mean BPS - no power control I
2
5.5 5.0 1 o - ~
Figure 4.4: The mean BER and BPS AQAM performance employing the threshold-based power con-
trol scheme for different dynamic ranges n, which was designed according to Table 4.1 in order to achieve a low mean BER The switching threshold levels were set to 11 = -mdB,
12 = 8dB, 13 = 14dB, 14 = 24dB and l 1 = 5dB, 12 = 8dB, 13 = 14dB, 14 = 24dB for the non-blocking and blocking schemes, respectively Perfect channel envelope inversion was also assumed in this narrow-band channel environment at the receiver
In this scenario, threshold-based power control was utilized, in order to increase the mean BPS performance This was achieved by increasing the transmitted power within the power control zone, whenever the expected received instantaneous power was below a certain thresh- old level This resulted in the employment of a higher-order modulation mode, which in- creased the mean BPS performance at the expense of the mean BER From this criterion,
Trang 114.2 POWER CONTROL ASSISTED ADAPTIVE MODULATION 91
"
Channel SNR(dB) AQAM(5, 8, 14,20)dB
(b) Relative frequency of power control utilization with and without transmission blocking
Figure 4.5: Relative frequency of power control and AQAM mode switching using the threshold-based
power control scheme for different dynamic ranges IC, which was designed according to Table 4.1 in order to achieve a low mean BER The switching threshold levels were set to
11 = - m d B , 12 = 8dB, 13 = 14dB, 14 = 24dB and 11 = 5dB, 12 = 8dB, 13 = 14dB,
14 = 24dB for the non-blocking and blocking schemes, respectively and perfect channel envelope inversion was assumed at the receiver
Trang 1292 CHAPTER 4 ADAPTIVE MODULATION
Previous
modulation:
64QAM 16QAM
4QAM BPSK
Table 4.2: The AQAM transition table, which was designed to achieve a high mean BPS The previous
modulation mode and the expected SNR level was stated and used in order to select the
present modulation mode The notations t and I, indicated the powering-up and powering-
down mode, respectively and K represented the maximum dynamic range of the threshold-
based power control scheme
the AQAM mode transition table can be formulated as shown in Table 4.2 The main differ-
ence between the transition tables shown in Figure 4.2 and Figure 4 l is that the powering-up
mode is used repeatedly for increasing the transmitted power, in order to utilize a higher-order
modulation mode, hence improving the mean BPS performance
The mean BER and BPS results are shown in Figures 4.6(a) and 4.6(b) for the blocking
and non-blocking AQAM schemes, respectively The effects of different dynamic ranges K ,
was also highlighted in these figures As expected, due to the BPS maximisation criterion
and the resulting power control methodology, the mean BPS performance improved slightly
at the expense of a higher mean BER The increase in the power control’s dynamic range
K , also improved the mean BPS performance, as depicted in Figures 4.6(a) and 4.6(b) This
trend can be explained similarly to the trends in Section 4.2.1, where a wider power control
zone resulted in a more frequent utilization of the threshold-based power control scheme,
hence increasing the mean BPS performance This increase in the power control utilization
frequency was evidenced by the results shown in Figures 4.7(b) The relative switching
frequency also displayed the same characteristics as those observed in Section 4.2.1, where
the utilization frequency for different dynamic ranges K , was approximately the same as
shown in Figure 4.7(a) The arguments of Section 4.2.1 can also be applied here
In summary, the employment of threshold-based power control in order to improve the
mean BPS performance resulted in the degradation of the mean BER, while the switching
frequency was more or less unchanged, when compared to the conventional AQAM scheme
In the next section, the utilization of the power control scheme is investigated in order to
reduce the switching frequency and its effects on the mean BER and mean BPS performance
Trang 134.2 POWER CONTROL ASSISTED ADAPTIVE MODULATION 93
* Mean BER - no power control
Figure 4.6: The mean BER and BPS AQAM performance employing the threshold-based power con-
trol scheme for different dynamic ranges K The threshold-based power control scheme
was designed according to Table 4.2 in order to achieve a high mean BPS The switching thresholds were set to 11 = -mdB, 12 = 8dB, 13 = 14dB, l4 = 24dB and l1 = 5dB,
12 = 8dB, 13 = 14dB, 14 = 24dB for the non-blocking and blocking schemes, respec- tively Perfect channel envelope inversion was also assumed in this narrow-band channel environment at the receiver
are observed
4.2.3 Threshold-based Power Control Designed for
Minimum Switching Utilization
In optimising the AQAM scheme for a lower switching frequency, the threshold-based power control scheme was designed to maintain the previous employed modulation mode if the short term SNR was within the power control zone and the previous modulation mode was a legit- imate one in the power control zone As a result of this criterion, the threshold-based power control scheme attempted to reduce the switching frequency The corresponding modulation mode transition table was formulated, which is shown in Table 4.3, utilizing this criterion
In this table, the powering-up and powering-down mode was used appropriately to ensure
that the modulation mode remained unchanged, whenever possible This was different from Table 4.1, where the powering-down mode was used exclusively for reducing the mean BER
Trang 1494 CHAPTER 4 ADAPTIVE MODULATION
I
"
Channel SNR(dB) AQAM(- 03, 8, 14,20)dB
Channel SNR(dB) AQAM(5,8, 14,20)dB
(a) Relative frequency of modulation mode switching with and without transmission blocking
Channel SNR(dB)
L
AQAM(5, 8, 14,20)dB
(b) Relative frequency of power control utilization with and without transmission blocking
Figure 4.7: Relative frequency of power control and AQAM mode switching using the proposed
threshold-based power control scheme for different dynamic ranges K , which was designed according to Table 4.2 to achieve a high mean BPS, The switching thresholds were set to
11 = - o d B , 12 = 8dB, 13 = 14dB, 14 = 24dB and 11 = 5dB, 12 = 8dB, l:$ = 14dB,
14 = 24dB for the non-blocking and blocking schemes, respectively and pcrfect channel envelope inversion was assumed at the receiver
Trang 154.2 POWER CONTROL ASSISTED ADAPTIVE MODULATION 95
Previous
modulation:
64QAM 16QAM 4QAM BPSK
N o T X
N o T X BPSK T N o T X N o T X No TX
Table 4.3: The AQAM transition table, which was designed to achieve a low switching utilization The
previous modulation mode and the expected SNR level was stated and used to choose the present modulation mode The notations and 1 indicated the powering-up and powering- down mode, respectively and IC represented the maximum dynamic range of the threshold- based power control scheme
and from Table 4.2, where only the powering-up mode was applied in order to increase the mean BPS performance As before, the performance of this power control scheme is anal- ysed in terms of its associated BER, BPS, switching utilization and power control utilization frequency
The associated mean BER and mean BPS performance results are shown in Figures 4.8(a) and 4.8(b) for the transmission blocking and non-blocking cases, respectively The perfor- mance results were compared to the conventional AQAM scheme without power control,
where there is a slight degradation in the mean BER at average channel SNRs in excess of 20dB for different dynamic ranges K At low average channel SNRs the utilization of the lower-order modulation modes was dominant, which enabled the power control scheme to
minimise the switching utilization frequency without degrading the mean BER or mean BPS performance However, at average channel SNRs in excess of 20 dB, the higher-order mod- ulation modes were selected more frequently Furthermore, the utilization of these higher- order modulation modes was maintained by the threshold-based power control regime, which was associated with a higher probability of errors This led to the slight degradation in the mean BER at average channel SNRs in excess of 20dB, as evidenced by Figures 4.8(a) and 4.8(b) As expected, under the switching frequency minimization regime, the switching
utilization decreased with increasing power control dynamic ranges, when compared to the conventional AQAM scheme This was evident in Figure 4.9(a), where there was a switch- ing utilization reduction of approximately 15%, when comparing the switching utilization
of the conventional AQAM scheme and the power control assisted AQAM scheme, which employed a dynamic range of = 1.5dB Finally, the power control utilization frequency,
Trang 1696 CHAPTER 4 ADAPTIVE MODULATION
(a) AQAM without Transmission blocking (b) AQAM with Transmission blocking
Figure 4.8: The mean BER and BPS AQAM performance employing the threshold-based power con-
trol scheme for different dynamic ranges K The threshold-based power control scheme was designed according to Table 4.3 in order to achieve a low switching utilization The
switching thresholds were set to 11 = -codB, l 2 = 8dB, Z3 = 14dB, 14 = 24dB and
11 = 5dB, 22 = 8dB, 13 = 14dB, 24 = 24dB for the non-blocking and blocking schemes, respectively Perfect channel envelope inversion was also assumed in this narrow-band channel environment at the receiver
which is depicted in Figure 4.9(b), increased as the dynamic range of the power control zone was increased This can be explained using the same arguments as in Section 4.2.1
The employment of threshold-based power control scheme according to Table 4.3 re- sulted in a reduced switching utilization frequency and a slight degradation of the mean BER
at high SNRs, when compared to the conventional AQAM scheme Let us now summarize the performance aspects of the different threshold-based power control schemes that we have discussed so far The summary of these performances is displayed in Table 4.4, where the per- formance was quantified by the mean BER, BPS, switching utilization frequency and power utilization frequency at a channel SNR of 20dB for the AQAM scheme with transmission
Trang 174.2 POWER CONTROL ASSISTED ADAPTIVE MODULATION 97
AQAM(- CO, 8, 14,20)dB AQAM(5, 8, 14,20)dB
(a) Relative frequency of modulation mode switching with and without transmission blocking
(b) Relative frequency of power control utilization performance with and without transmission blocking
Figure 4.9: Relative frequency of power control and modulation mode switching using the threshold-
based power control scheme for different dynamic ranges K , which was designed according
to Table 4.3 in order to achieve a low switching utilization The switching thresholds were set to 11 = -codB, 12 = 8dB, 1 3 = 14dB, 14 = 24dB and 11 = 5dB, 12 = 8dB,
13 = 14dB, 14 = 24dB for the non-blocking and blocking schemes, respectively and
perfect channel envelope inversion was assumed at the receiver
Trang 1898 CHAPTER 4 ADAPTIVE MODULATION
Threshold Level Power Control for lower Mean BER Mean BER Power Mean BPS Switching Util.(%) Util.(%)
K = 0.5
23.9 66.6 3.78
4.96 x 10-4
K = 1.5
15.7 66.5 3.93
8.06 x 10-4
K = 1.0
7.6 66.1
4.07 1.25 X 10V3
3.62 x lop3
K = 1.5
14.6 62.7 4.49 3.09
x l o p 3
K = 1.0
7.4 64.1 4.35 2.49 X 10-’
Threshold-based Power Control for lower Switching Utilization
Mean BER Mean BPS Switching Util.(%) Power Util.(%)
K = 0.5
9.6 55.6 4.23 2.0 x 10-3
K = 1.0
4.8 60.4 4.22
1.9 x 10-3
Conventional x l o p 3 4.22 1.84 65.4
Table 4.4: The mean BER, mean BPS, switching utilization and power control utilization relative fre-
quency for the three different threshold-based power control designs of Tables 4.1, 4.2 and 4.3 for different dynamic ranges IC These performance measures were recorded at an aver-
age channel SNR of 20dB, where the switching threshold levels for this transmission block- ing AQAM scheme were set to ZI = 5dB, l 2 = 8dB, l 3 = 14dB, 14 = 24dB These performances were compared to that of the conventional AQAM scheme, which did not
utilize any power control scheme
performance was improved at the expense of the mean BER performance The scheme was also utilized in order to lower the modulation switching utilization frequency, while main- taining the mean BER and BPS performance However, these performance enhancements resulted in more frequent utilization of power control, which was an additional system over- head The performance of the threshold-based power control scheme in all scenarios was influenced by the dynamic range K , where, the scheme achieved its aim better for a higher value of K , as evidenced by the results in Table 4.4 However, the disadvantage in doing this was that the power control utilization frequency increased
In the next section, we shall concentrate on the application of AQAM in a wideband channel environment, where we can still use the principles of AQAM developed for a narrow- band environment which we have discussed in this section
Trang 194.3 ADAPTIVE MODULATION AND EQUALIZATION IN A WIDEBAND ENVIRONMENT 99
4.3 Adaptive Modulation and Equalization in a
Wideband Fading Environment
In the last section the concepts of AQAM were discussed in the context of a narrow-band fading channel environment [145] In this section we will extend those concepts in order
to apply AQAM in a wideband fading channel environment, where equalization plays an important role as evidenced by the results presented in Chapters 2 and 3 Consequently,
when applying AQAM in a wideband environment, the joint optimisation of the equalizer and the AQAM scheme is necessitated
In the narrow-band channel environment the quality of the channel was determined by the short-term SNR of the received burst, which was then used as a metric in order to invoke the appropriate modulation mode at the transmitter, based on a list of switching threshold levels
l , [145] However, in a wideband environment this metric is not applicable as an estimate
of the quality of the channel, where the existence of dispersive multi-path components in the wideband channel produces not only power attenuation of the transmission burst, but also intersymbol interference, as discussed in Section 2.2 Consequently, the metric used to estimate the channel’s quality has to be redefined
The wideband channel will introduce transmission degradation in terms of signal power fluctuations and intersymbol interference as a result of its dispersive fading multi-path na- ture Thus the criterion used to switch the modulation modes must incorporate these two effects of the wideband channel Accordingly, in this wideband channel environment the channel-induced degradation is combated not only by the employment of AQAM but also by equalization In following this line of thought, we can formulate a two-step methodology in mitigating the effects of the dispersive wideband channel In the first step, the equalization process will eliminate most of the intersymbol interference based on a CIR estimate and con- sequently, the signal to noise plus residual interference ratio at the output of the equalizer is calculated based on Equations 2.58, 2.59,2.60 and 2.61 of Section 2.4, termed as the pseudo- SNR output This pseudo-SNR at the output of the equalizer is then used as a metric to switch the modulation modes By utilizing this pseudo-SNR, we are ensuring that the system per- formance is optimized by jointly employing equalization and AQAM techniques in order to mitigate the effects of the dispersive multi-path fading channel
In the forthcoming sections, the proposed wideband AQAM and equalization scheme is explored with the aim of characterizing its upper bound performance The optimisation of the switching threshold levels is also analysed, in order to achieve a certain target mean BER and BPS performance Finally, the performance of this wideband AQAM scheme and that of its individual fixed modulation modes is compared in terms of their transmission throughput Before proceeding further, let us state the assumptions used in the employment of this scheme
in a wideband channel environment
Trang 20100 CHAPTER 4 ADAPTIVE MODULATION
3 At the receiver, perfect channel compensation is applied in order to achieve the upper bound performance In Chapters 2 and 3 , we have noted the performance degradation due to incorrect CIR estimation for fixed-mode modulation schemes However, this degradation is neglected here with the aim of achieving the upper-bound performance Nevertheless CIR estimation techniques will be invoked and studied in a wideband AQAM scheme in Chapter 7
4 The receiver assumed perfect knowledge of the modulation mode used in its received transmission burst In reality, some form of modem mode control signalling must be employed to convey the modulation mode used to the receiver [21,37] As a result of the dispersive channel, it is likely that these control symbols may become corrupted sufficiently for the receiver to make an erroneous decision on the modulation mode, which in turn might corrupt the demodulation process However in Section 5.6 we will attempt to detect the modulation mode by exploiting the extra information provided by the channel decoder at the receiver
5 The equalizer employed in this scheme is the DFE, where it is assumed that error prop- agation can be neglected This will simplify the calculation of the associated numerical upper bound performance Nevertheless, the impact of error propagation is included and studied at a later stage in Section 7.1 for the wideband AQAM scheme
6 The residual IS1 at the output of the DFE is assumed to be Gaussian distributed for the purpose of mean BER calculations This assumption will be justified with the aid of an experiment at a later stage
Above, we have outlined and justified the assumptions required, in order to achieve the upper bound performance of this wideband AQAM and equalization scheme Let us now concentrate our attention on the methodology of this scheme
The system schematic of the wideband AQAM and equalization scheme is depicted in Figure 4.10 At the receiver the channel quality is estimated, which is then used to calculate the DFE coefficients via the DFE coefficient estimator block of Figure 4.10 by solving Equations 2.54
Trang 214.3 ADAPTIVE MODULATION AND EOUALIZATION IN A WIDEBAND ENVIRONMENT 101
Figure 4.10: Schematic system overview of the wideband AQAM and equalization scheme
and 2.55 of Section 2.3.4 Subsequently, the coefficients are used to equalize the corrupted received signal In addition to that, both the CIR estimate and the DFE coefficients are utilized for computing the pseudo-SNR at the output of the DFE The calculated pseudo-SNR is then compared against a set of optimized switching threshold levels t,, stored in a look-up table Consequently, a modulation mode is selected for the next transmission burst, assuming reciprocity of the uplink and downlink TDD slots, where there is a close correlation between the uplink and downlink CIR estimates This implies that the reciprocity of the pseudo-SNR for the uplink and downlink transmission can be exploited, in order to set the next modulation mode at the transmitter The modulation modes that are utilized in this scheme are BPSK, 4QAM, I6QAM, 64QAM and a no transmission (NO TX) mode, which were also used by Torrance [3 1,1451 in a narrow-band channel environment The methodology of switching the modulation modes is similar to that of Torrance [31,145], but instead of using the short term transmission SNR as a switching criterion, the pseudo-SNR at the output of the DFE Y D F E ,
where t,, R = 1 4 are the pseudo-SNR threshold levels, which are set according to the
required mean BER and BPS throughput Let us now investigate the validity of using the
pseudo-SNR at the output of the DFE Y D F E , as a switching criterion in this wideband AQAM and equalization scheme
4.3.3 The Output Pseudo Signal to Noise Ratio of the Decision Feedback
Equalizer
In the last section, we introduced the concept of using the pseudo-SNR at the output of the DFE as our channel quality metric for switching the modulation modes and hence the appli- cability of this metric in applying AQAM is investigated in a wideband channel environment