08-10-2012, 01:32 PM
Analysis of Impact on Handset Transmitter Design of the High-speed Data Requirements in the IS-95-B CDMA Wireless Standard
Analysis of Impact.pdf (Size: 327.85 KB / Downloads: 23)
Abstract
As part of the recent evolution of the IS-95 CDMA wireless communications towards
third-generation technology, high-speed data requirements were defined for both the
uplink and downlink in the recently proposed IS-95-B revision. Although the requirements
for the downlink are straightforward, the uplink requirements have considerably
larger impact on the handset transmitter and base station receiver design. In this
paper, the impacts on the transmitter are examined, jkom baseband to RF frequencies.
In particular, waveform quality and emissions are examined through high-level simulation
to determine the tradeoffs involved inimplementation of a handset design which
can support IS-95-B high speed data requirements for the uplink in the 1.g) GHz band.
Introduction
The IS-95 standard poses many challenges for
implementation, particularly in the design of a
proper transmitter. The required design must be
linear over a large dynamic range (nearly 100 dB)
yet still be designed efficiently SO as to compete
with other existing wireless standards. Therefore,
careful system analysis is oftentimes necessary
prior to implementing enhancements to the
existing IS-95 functionality.
Recently, the IS-95 standard has been expanded
to include high-speed data functionality in both
the uplink and downlink [2]. The implementation
for multiple user data channels, i.e. traffic
channels, is straightforward in the downlink. In
this case, the mobile user is already required to
demodulate multiple data channels, each defined
by a unique Walsh code sequence, in order to establish
and maintain a communications link. The
advantage of Walsh code sequences is that due to
their mutual orthogonality, one particular channel
may be extracted while suppressing all other channels
at the same time simply by modulating the
incoming data with the desired Walsh sequence
Modeling and Simulations
The modeling depicted in Figure 1 and Figure
2 was performed so as to properly simulate
a CDMA system using MATLAB. Random bits
were generated and passed through the baseband
filters, which were modeled with the FIR coefficients
quantized to 9 bits. The 8-bit DAC’s on
both the I and Q channels were modeled with
.9 bits of differential nonlinearity (DNL) and 2
bits of integral nonlinearity (INL). It was also assumed
that 32 bits at the output of each DAC
were reserved for possible DC-offset correction,
and as a result only 224 levels were available at
each DAC. Each DAC’s nominal output ranged
from -.75 volts to -75 volts. These DAC’s were
followed by image-reject filters, modeled as thirdorder
Butterworth type.
It is important to take note of the scaling of
the FIR filter output samples, as this is how one
maximizes the use of the available headroom on
the DAC. Moreover, one must chose scaling factors
that are sufficiently robust regardless of the
number of data channels being transmitted; this
requires some experimentation.
Conclusions
Based on preliminary systems-level simulation
results, it can be concluded that there is a considerable
cost associated with implementation of IS-
95-B high speed data requirernentis for the uplink
if one chooses to support the malximum bitrate
provided by 8 data channels. The degradation
in performance due to the increase in dynamic
range of the data may be offset if one increases
the resolution of the I and Q DAC’s. Given that
DAC’S required are high-freq1uenc.y (on the order
of 5 Msps), this requirement can result in an expensive
design revision for existing IS-95 mobile
handset architectures