Digital Signal Processing Reference
In-Depth Information
Based on the bitrate, communication performance is determined by the Bit Error
Rate (BER) at the receiver. When transmitter non-linearity is considered, the BER
is expressed as a function of the above introduced SINAD. The SINAD is written
as a function of the power amplifier back-off, given output power
P
Tx
and time and
frequency varying channel attenuation
A(t, f )
as:
P
Tx
×
A(t, f )
SINAD
=
N
f
,
(6.2)
×
+
×
×
A(t, f )
D
i
(b)
kT
W
P
Tx
η
PA
(b)
,
P
PA
=
(6.3)
where the constants
k
,
T
,
W
and
N
f
are the Boltzmann constant, working temper-
ature, channel bandwidth and noise figure of the receiver respectively. The relation
between the power amplifier back-off
b
and the distortion has been characterized
empirically for the Microsemi LX5506 [81] 802.11a PA in Fig.
6.4
. The effective
PA power consumption (
P
PA
) can be expressed as the ratio of the transmit power
(
P
Tx
) to the PA efficiency (
η
PA
) that is related to
b
by an empirical law fitted on
measurements (Eq.
6.3
).
Next, at MAC layer, we can add the number of retransmissions as control knob.
These knobs are practical and sufficient to illustrate the potential of the proposed
methods.
6.2.2 The Varying Context
System State
(
S
i,m
): In a wireless environment, with e.g. VBR video traffic, the sys-
tem state scenarios used in this design case are the current channel state and appli-
cation frame size. We study the impact of both constant bitrate (CBR) and variable
bitrate (VBR) traffic. For VBR traffic we employ MPEG-4 encoded video traces [82,
83] with peak-to-mean frame sizes ranging from 3 to 20. All fragmentation is done
at the link layer and we use UDP over IP. As the maximum frame size is assumed to
be within the practical limit of 50 fragments, we construct Cost-Resource-Quality
curves for 1, 2, 3, 4, 5, 10, 20, 30, 40, 50 fragments/frame and interpolate for inter-
mediate values. The application layer frame size is translated to lower layer Queue
Size and fragment size to facilitate state monitoring and calibration. As a result, no
additional measurements are needed to model traffic requirements, which can be
fully captured in the mapping.
The indoor wireless channel suffers from frequency-selective fading that is time-
varying due to movements of the users or obstacles in its environment. This varying
fading results in varying PER as function of the current transceiver settings. We
use a frequency selective and time varying channel model to compute this PER for
all transceiver settings. An indoor channel model based on HIPERLAN/2 [84]was
used for a terminal moving uniformly at speeds between 0 to 5.2 km/h (walking
speed). Experiments for indoor environments [83] have found the Doppler spread to
be approximately 6 Hz at 5.25 GHz center frequency and 3 Hz at the 2.4 GHz center
