Digital Signal Processing Reference
In-Depth Information
Instead of this, the receiver passively relies on a combination of statistics and
the delay spread in the channel. This strategy has already been extensively dis-
cussed in Section 4.4: combining the energy of multiple resolved multipaths
- in this case the energy of the parallel receive branches - greatly improves
the reliability of the wireless link. All this without the need to worry about ev-
ery single 'pulse'-thing that arrives at the antenna or finding the optimal offset
of the receive window. The only necessary precondition for this approach to
work is that the different receiver branches experience independently fading
multipaths, which is the reason why the transmit window should be as short as
possible.
The reader may wonder whether the pulse-based radio system with multiple
receive modules actually offers any advantage over the ofdm-based system
with a comparable sample rate. After all, except for the fact that data is being
processed by multiple receive branches in parallel, the overall throughput of
raw data symbols is not less in a pulse-based receiver than it is for the ofdm-
system.
Parallel processing
There are many benefits of parallel processing, though. A parallel architec-
ture reduces the power consumption while maintaining the same throughput
as the single-path ofdm system. Also, the system of parallel receive modules
allows to save additional power by disabling some of the branches in case of a
sufficient signal strength in the remaining paths. Depending on the actual per-
formance requirements, the pulse-based wideband radio system allows flexible
scaling, without making changes to the specifications of the transmitted signal:
the designer of a cheap, battery-powered system could decide to reduce the
multipath resolvability of the receiver in exchange for a reduced level of par-
allelism. In addition, probably the most underestimated energy saver of the
pulse-based radio system is that in-band interferer power is effectively divided
over the parallel paths, resulting in a drastic reduction of the linearity require-
ments (iip
3
) in each of the front-end modules in the receiver. For example,
if the on-time of the receive window is one tenth of the total period between
successive pulses, in-band interferer power is suppressed by 10 dB. In contrast,
the front-end of the ofdm-based system is exposed to the full interferer power.
The parallelism of the pulse-based radio architecture also allows the use of
more sophisticated algorithms to improve the performance of the receiver, even
without the cost of extra hardware. Refer back to the example above, where the
interval between two pulses is uniformly divided over 10 receive units. Instead
of trying to cover the complete interval, it is also possible to eliminate half of
the parallel units using an alternating pattern, as shown in Figure 5.11. If the
delay spread of the channel is larger than the duration of two receive windows,
