Professional Documents
Culture Documents
ir
http://www.itrans24.com/landing1.html
u m
Searching Rake Receiver
assigned a group of Rake fingers with equal spacing
around the detected code phase. Allocating the Rake
fingers with spacing less than one chip period can re- Re-allocate
duce overhead for delay tracking. The grouped Rake
finger allocation principle is compared to single Rake
finger allocation. Allocating three Rake fingers with Fig. 1. Block diagram of the acquisition and receiver
T,f 2 spacing gives a 0.2-0.4 dB gain compared to sin- mechanism.
gle Rake finger assignment with the same number of
Rake and tracking fingers.
in groups. This paper analyzes the performance
I. Introduction of a Rake receiver with Rake fingers being allo-
cated in groups.
After code phase acquisition has been estab- The organization of the paper is the following.
lished the Rake finger must be associated with The channel model used for evaluating the Rake
the detected code phase. The detected code finger management is described in section 11.
phase corresponds to a multi-path delay in the
Traditional Rake finger allocation is described
radio channel. The delays of the channel multi- in section 111. The proposed grouped Rake fin-
paths are none stationary, therefore it is nec- ger allocation method is analyzed in section IV.
essary to continuously look for new multi-path The performance of the two allocation scheme is
components in the channel. compared in section V. Concluding remarks are
The multi-paths are then allocated to the Rake given in section VI.
receiver for demodulation. This coarse synchro-
nization for each Rake finger must be fine syn- 11. Radio Channel Measurements
chronized by the tracking mechanism in the re-
ceiver. The multi-paths allocated to the Rake It is desirable to test the acquisition and
receiver will disappear as the mobile moves and Rake finger management algorithms in a real
the delay profile of the channel changes. These transceiver. Since a real time UMTS test-bed
fingers are then de-allocated from the Rake re- was unavailable in the project, measurement
ceiver and new multi-paths can be allocated from data can be used to simulate receiver perfor-
the searching mechanism. The inter working be- mance off-line. Such measurement data were
tween searching and the Rake receiver is shown available from CPK at Aalborg University [l].
in figure 1. The measurements were performed in Aarhus,
The multi-paths are detected by the code phase Denmark and in Stockholm, Sweden. The sys-
acquisition device and the strongest multi-paths tem was designed for uplink transmission. How-
are each allocated a Rake finger. Rake fingers ever, the same channel properties can be ap-
can be grouped as multi-paths appear and a sin- plied to the downlink channel. The transmit-
gle tracking mechanism can be shared among the ter is placed in the measurement vehicle. The
Rake fingers. Rake fingers can also be allocated measurement vehicle follows a given path with a
Fig. 2. Cumulative density function of the channel The finger allocation algorithm is the connection
delay spread measured in Aarhus and Stockholm. between the Rake receiver and the search fingers.
After the search fingers find the strongest multi-
been estimated from recorded samples prior to path components they have to be allocated to
the impulse response. The analysis uses -20 dB the Rake receiver for demodulation. The multi-
noise floor and any power below that threshold is path components change as the mobile station
regarded as zero. The channel rms delay spread moves around and the current multi-path com-
is shown in figure 2. The median channel rms ponents in the Rake receiver have to be removed
delay spread for Aarhus and Stockholm is 1.3 ps or de-allocated. The mechanism of allocating
and 2 ps respectively. multi-paths to and from the Rake receiver is de-
Searcher
priority list "0 2 4 6 B 10 12 14 16 18 20
Number of Rake fingers required
Path 11 Next
Rake allocation
Fig. 5 . Cumulative density function of the number
Finger 17
of Rake fingers needed to ensure 70-90% of the
Finger 13 Finaer I n
Path 16 Finger /4
received power as indicated in the plot.
Path 17 Finger /6
used to distinguish the impulses from noise, thus
Fig. 4. An example of allocating Rake fingers to all samples of the impulse response with power
measured multi-path components. less that -10 dB of the strongest peak. The total
power is calculated as the sum of all power peaks
are ordered from largest power to the lowest in in the impulse response
a searcher priority list. This is the output from
the code phase acquisition. The strongest multi-
path from the searcher is allocated to available
Rake fingers. Even though a Rake finger is allo- The strongest peaks are identified, and the cu-
cated to a multi-path component, it may not be mulative sum of these peak powers are calculated
included in the Rake combining due to fading. as n
This will ensure better performance if the path
with low instantaneously SNR is excluded in the
Rake finger combining.
where P, are the strongest power peaks in de-
B. Number of Rake fingers scending order. The distribution of the receiver
power utilized is the ratio
The number of Rake fingers is very important
for an implementation of the Rake receiver. The Pn
Pused =- (3)
baseband chip should be designed with a number Ptotal '
of Rake fingers suitable for all environments. For which is shown in figure 5 for Pus& between 70%
low dispersive environments the unused Rake and 90% of the total received power.
fingers should be turned off. The requirement to The typical urban environment in Aarhus re-
the number of Rake fingers has been analyzed quires 4 to 6 Rake fingers depending on the
in [2] for three different environments in Ade- power ratio used in the comparison. This is in
laide, Melbourne, and Sydney in Australia. It good agreement with results in [2]. However for
was found that in a system with chip rate 4 Mcps the bad urban environment in Stockholm 10 to
four Rake fingers would utilize 80% of the total 15 Rake fingers is more realistic for optimum per-
received power in 95% of all cases. formance.
02000 IEEE.
0-7803-57 I8-3/00/$I0.00 89 VTC2000
Downloaded from http://iranpaper.ir
http://www.itrans24.com/landing1.html
IV. Grouped Rake Finger Allocation of consecutive taps is in the order of 0.5 - 1.0 ps
corresponding to 2 - 4 chip periods. The chip
A Rake receiver consists of a bank of correlators.
rate of UNITS is so high that often more than one
Each of the correlators is used to detect sepa-
Rake finger is necessary to capture the energy
rately one of the L strongest multi-path compo-
from the main impulse response tap. By allo-
nents. Each of the decorrelated multi-paths is
cating multiple Rake finger with chip separation
combined into one signal before detection. For gives a great overhead for tracking. It is there-
the operation of the Rake receiver it is neces-
fore suggested to allocate some of the Rake fin-
sary to identify and track the major multi-path
gers group wise with a single tracking mechanism
components as well as to estimate their rela-
associated with the group of Rake fingers. The
tive delays, amplitudes, and phases. The first
original method suggested by Price and Green
Rake receiver described in 1958 by Price and
would require a high number of Rake fingers in
Green [3] used a fixed number of Rake fingers
order to capture all the power from a time dis-
each equally separated with T, seconds equal to
persive channel including including signal power
the chip interval. The usability ofthis finger al-
from multiple base station in soft handoff.
location method works well for low bandwidths
This method splits the received channel impulse
where a few Rake fingers will cover large excess
response into a few separate bins. Each bin is as-
delays. However, for high chip rates and systems
signed a set of Rake fingers and a single tracking
receiving from multiple base station it may not
mechanism. An example of the grouped Rake
be flexible enough to have all the Rake fingers
finger assignment is shown in figure 7. Each
separated by T, seconds.
In todays mobile communication, a searcher
looks for each individual multi-path component
and then allocates a single Rake finger, each with
a tracking mechanism [4]. This finger alloca-
tion scheme works well for mobile communica-
tion systems receiving from multiple base sta- Fig. 7. Grouped Rake finger assignment technique.
tion simultaneously. The channel impulse re-
sponses obtained from Aarhus and Stockholm Rake finger group has assigned a single tracking
have been analyzed for the distribution of the mechanism. The number of Rake fingers in each
multi-path components and which Rake finger group depends on the instantaneous radio chan-
allocation method to chose. The cumulative den- nel impulse response. The allocation of Rake
sity function for the width of consecutive impulse fingers in groups can be done in many differ-
response taps in both Aarhus and Stockholm is ent ways depending on the available number of
shown in figure 6. It can be seen that the width Rake and tracking fingers. The main advantage
of allocating Rake fingers in groups is to reduce
the overhead for tracking. The following sections
1
will analyze different allocation methods based
09
on the grouped Rake finger concept.
08
For simplicity the analyzed grouped Rake finger
allocation algorithm uses equal size Rake finger
groups. This gives an over estimate of the re-
quired number of Rake fingers to be used in this
scenario. Using a separation between Rake fin-
gers less than the chip period, T, reduces the
overhead for tracking fingers. By allocating the
Rake fingers in groups of three spaced less than
0 05 1 15 2 25 3 35
Wdm of consecultve impulse response laps Ips]
4 45 5 Tc the two outer Rake fingers can be used for
delay tracking as well as descrambling. This is
Fig. 6. Cumulative density function of the width of the same architecture as the ideal Early-late gate
consecutive impulse response taps measured in tracker with the exception that all three fingers
Aarhus and Stockholm. are used in the Rake combining. Results com-
V. Receiver Performance
The performance of the Rake receiver is highly
dependent on detection and allocation of multi- w
U