The RAKE Receive Process

The signal, transmitted by the Node B or handset, will usually travel along several different
paths to reach the receiver. This is due to the reflective and refractive surfaces
that are encountered by the propagating signal. Because the multiple paths have different
lengths, the transmitted signal has different arrival times (phases) at the receive
antenna; in other words, the longer the path the greater the delay. 1G and 2G cellular
technologies used techniques to select the strongest path for demodulation and processing.
Spread spectrum technology, with its carrier time/phase recognition technique,
is able to recover the signal energy from these multiple paths and combine it to
yield a stronger signal.
Data signal energy is recovered in the spread spectrum process by multiplying synchronously,
or despreading, the received RF with an exact copy of the code sequence
that was used to spread it in the transmitter. Since there are several time-delayed versions
of the received signal, the signal is applied simultaneously to a number of synchronous
receivers, and if each receiver can be allocated to a separate multipath signal,
there will be separate, despread, time-delayed recovered data streams. The data
streams can be equalized in time (phase) and combined to produce a single output.
This is the RAKE receiver.
To identify accurately the signal phase, the SCH is used. As already described, the
received RF containing the SCH is applied to a 256-chip matched filter. This may
be analog or sampled digitized IF. Multiple delayed versions of the same signal will
produce multiple energy spikes at the output. Each spike defines the start of each
delayed slot. (It is the same slot—the spikes define the multiple delays of the one slot.)

Each spike is used as a timing reference for each RAKE receive correlator. The
despreading code generator can be adjusted in phase by adjusting the phase of its
clock. The clock is generated by an NCO—a digital waveform generator—that can be
synchronized to a matched filter spike, that is, the SCH phase (see Figure 3.16).
Because the received signal has been processed with both scrambling and spreading
codes, the code generators and correctors will generate scrambling codes to descramble
(not despread) the signal and then OVSF codes to despread the signal. This process
is done in parallel by multiple RAKE receivers or fingers. So now, each multipath echo
has been despread but each finger correlator output is nonaligned in time. Part of the
DPCCH, carried as part of the user-dedicated, or unique, channel is the pilot code.
The known format of the pilot code bits enables the receiver to estimate the path
characteristic—phase and attenuation. The result of this analysis is used to drive a
phase rotator (one per RAKE finger) to rotate the phase of the signal of each finger to a
common alignment. So, now we have multiple I and Q despread bit streams aligned in
phase but at time-delayed intervals.
The last stage within each finger is to equalize the path delays, again using the
matched filter information. Once the phase has been aligned, the delay has been
aligned, and the various signal amplitudes have been weighted, the recovered energy
of interest can be combined.

Path combining can be implemented in one of two ways. The simpler combining
process uses equal gain; that is, the signal energy of each path is taken as received and,
after phase and delay correction, is combined without any further weighting. Maximal
ratio combining takes the received path signal amplitudes and weights the multipath
signals by adding additional energy that is proportional to their recovered SNR.
Although more complex, it does produce a consistently better composite signal quality.
The complex amplitude estimate must be averaged over a sufficiently long period
to obtain a mean value but not so long that the path (channel) characteristic changes
over this time, that is, the coherence time.