19.11.2014 Views

mohatta2015.pdf

signal processing from power amplifier operation control point of view

signal processing from power amplifier operation control point of view

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

60 ZERO-FORCING DECISION FEEDBACK EQUALIZATION<br />

The decision variable for si is then given by Z\ = —0.1j/i, which can be modeled as<br />

2i=si+0.1ni. (3.11)<br />

Assuming a noise power of 100, the resulting output SINR is 1.0, which is greater<br />

than the MF output SINR of 0.955. Thus, in this case, we expect the ZF DFE to<br />

perform better at high SNR (when decisions are mostly correct).<br />

We can also determine a lower bound on output SINR by assuming incorrect<br />

subtraction of ISI. When we subtract the incorrect value, we essentially double the<br />

value (e.g., +1 — (—1) = 2). The model for y\ becomes<br />

2/1 = -10s 1 +9(2s„) + ni. (3.12)<br />

The decision variable for si is then given by z\ = — O.lyi, which can be modeled as<br />

zi = s-i +1.8s()+0.1m. (3.13)<br />

The resulting output SINR is 0.236, which is much lower.<br />

Is zero-forcing the best strategy? While it eliminates ISI, it doesn't account for<br />

the loss of signal energy by ignoring the second copy of the symbol of interest in r 2 .<br />

It turns out that we can use future samples to recover some of this loss. However,<br />

then we can't entirely eliminate ISI. This will be explored in Chapter 5.<br />

In the general dispersive scenario, for s m , we form the decision variable<br />

and detect Sfc using<br />

The upper bound on output SINR is then<br />

ym=r m -dê m -i (3.14)<br />

s m = sign{cy m }. (3.15)<br />

SINR

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!