Note: Descriptions are shown in the official language in which they were submitted.
13~9~4~3
~ 5MR 511
APPARATUS AND METHOD FOR
HIGH-SPEED DETERMINATION OF
RECEIVED RF SIGNAL STRENGTH INDICATOR
FIELD OF THE INVENTION
The invention ralates to techniques for
measuring the amplitude of a radio frequency signal
and more particularly, to arrangement for
determining received signal strength indicator
(RSSI) in a cellular radiotelephone system.
,BRIEF DESCRIPTION OF THE DRAWINGS
Because the Background of the Invention
makes reference to Figure 1 o~ the dxawings, a
brief description of the drawing precedes the
Background of the Invention. For a better
understanding of the nature and objects of the
present invention, reference may be had by way of
example to the accompanying drawings, in which:
FIGURE 1 is a schematic illustration of a
simplified cellular radiotelephone communications
system;
FIGURE 2 is a detailed block diagram of
the presently prefsrred exemplary embodiment of an
RF received signal strength indicator system in
accordance with the present invention;
. , . , . ~ . .
. . ,, . -
- . . . .
' . ' '
'
~36:~9~3
- 2 - 45MR 511
FIGURES 3 and 4 are flowcharts of some of
the relevant control function steps performed by
the embodiment shown in FIGURE 2;
FIGURES 3(A) and 4(A) are flowcharts of
some of the relevant control function skeps
performed by a further presently preferred
embodiment of the present invention; and
FIGURES 5(A) and 5(B) are graphical
illustrations of parameters measured by the
embodiment shown in FIGURE 2 for two different
exemplary received RF signalsO
BAC~GROUND OF THE INVENTION
The basic structure and operation of a
cellular radiotelephone system has been disclosed
in a variety of publications. See, for example,
the January 1979 issue of The Bell SYstem Technical
Journal; and Specification EIA IS-3-B entitled
"Cellular System Mobile Station-Land 5tation
Compatibility Specification" (July, 1984,
Electronic Industries Association).
As is well known, the process called
"hand off" is a fundamental part of the cellular
radiotelephone scheme. A simplified cellular
radiotelephone system 10 is shown in FIGURE 1.
Cellular system 10 includes several fixed RF
transceiving station 12 each serving an associated
discrete geographical area ("cell") 14. A central
controller 16 supervises and controls the operation
of the fixed stations 12. As a mobile station 18
moves from a first "cell" (e.g., cell 14B) to a
second "cell" (e.g., 14C), the central controller
16 controls the fixed station 12B serving the first
cell 14B to discontinue handling the mobile
station's cell and controls the fixed station 12C
. " , . . .
'
~L3~9~L3
- 3 - 45MR 511
serving the second cell 14C to begin handling the
call (and also controls the mobile station to
retune to a frequency fixed station 12C operates
on). In this way, the mobile station 18 ~and its
call) is "handed off" to the cell receiving the
strongest signal from the mobile station. High
guality communications is thus maintained even
while mobile station 18 is moving from one cell to
another.
System 10 measures the RF signal strength
of transmissions of mobile station 18 at the
locations of fixed stations 12 in order to decide
when a hand off should occur. Decreased received
signal strength at a fixed station 12 indicates
that the mobile station 18 transmitting the signal
is nearing the edge of the cell 14 served by the
fixed station and is likely to need handing off to
a different c~ll. Signal strength measurements
performed by fixed stations 12 serving adjacent
cells are used to determine which cell the call
should be handed off to (the call is generally
handed off to the cell receiving the mobile station
transmission at the highest received signal
strength), thus maximizing communications quality
and reliability and minimizing the number of
hand-offs necessary.
When system design includes partitioned
cells (pie-shaped sectors, overlayed cells, etc.),
signal strength measurements at fixed stations may
also be used to determine which cell partition may
best serve particular mobile stations. Signal
strength measurements using mobile equipmant may be
used to verify the RF field strength pattern of
fixed station transmissions for purposes of
propagation analyses.
.
: . .
' . , ' ' ~
~309143
- 4 - 45MR 511
As will be appreciated, signal strength
measurements are very important in the design and
operation of cellular radiotelephone communications
systems, and are indeed an essential requirement of
cellular equipment design.
Every hand off in a cellular
radiotelephone system requires a number of signal
strength measurements. Since cellular systems
typically serve large numbers of mobile stations,
many signal strength measurements are required.
Moreover, because mobile stations are usually in
motion, the cellular system must respond very
rapidly to changes in received signal strength
(e.g., by handing off calls) to maintain acceptable
signal levels as mobile stations move from cell to
cell. There is therefore a great need for fast and
accurate received signal strength measuring
techniques.
RF signals transmitted by mobile radio
stations are subject to Rayleigh fading, as is well
known. Fades are of short duration and may be
twenty dB or more below the average received signal
strength level, making accurate and rapid signal
strength measurements difficult to obtain (a
measurement made during a deep fade is not
representative of the true average received signal
strength).
Prior art methods of overcoming this
difficulty include analog filtering (equivalent to
damping a meter movement so that it does not
respond to fast transients), and mathematical
averaging of a number of measured samples of
received signal strength. Such prior art
techniques require several measurements to be taken
. . ~ ~' ' ~ - .
~3~91~3
- 5 - 45MR 511
over a period of time large enough to mask the
effects of fading. The number of samples averaged
using the averaging technique must be great enough
so that measurements made during fades do not
unduly influence the resulting averaga.
Such prior art techniques suffer from at
least two disadvantages. First, the extended time
period required to obtain accurate measurements
using such technigues in is conflict with the
requirement that received signal strength
measurements must be made as rapidly as possible.
Second, it is often desirable to be able to measure
sudden changes in the average signal level, such as
when the mobile passes behind a large obstacle
which "shadows" the antenna. For a rapidly moving
vehicle, these changes may occur only a little more
slowly than the Rayleigh fading which it is
desirable to mask.
Roth the averaging and damping techniques
of the prior art tend to mask these rapid signal
strength changes along with the received signal
strength changes attributable to Rayleigh fading
phenomenon. As the "damping" (or the number of
samples being averaged) is increased to overcome
fading effe.cts, received signal strength
measurement becomes insensitive to other
~luctuations in received signal strength which it
may be helpful or desirable ~o measure. As a
result, the cellular system may respond too slowly
to changes in signal strength, allowing the mohile
station to receive unacceptable service quality and
perhaps even causing the loss of service. Even
more important, the excess time required for
measurement reduces the number of mobile stations
... ,,, . ,l
,. . .
1309~43
- 6 - ~5MR 511
which can be handled, i.e., additional equipment is
required to increase system capacity.
The technique disclosed in U.S. Patent
No. 4,549,311 to McLauyhlin (October 22, 1985)
measures the strength of a RF signal by sampling
the signal two or more times during a predetermined
time interval and selecting the sampled signal
strength having the largest magnitude. The method
taught by this McLaughlin patent is essentially a
digital implementation of a peak reading meter.
The McLaughlin technique always chooses the largest
of a plurality of samples (i.e., the peak received
signal strength), and therefore is insensitive to
received signal strength fluctuation attributable
to the values of other measurements in the sampling
interval. The technique will begin to detect rapid
changes in the average only when a new sampling
interval i5 obtained.
There is great need for an accurate
high-speed received signal strength measuring
technique which masks the effect of Rayleigh
fading, but which is sensitive to received signal
strength changes caused by effects other than
Rayleigh fading (e.g., obstacles in the signal5 transmission path of a moving mobile station).
SUMMARY OF THE INVENTION
The present invention provides a rapid
and accurate estimate of the average strength of a
radio frequèncy ~ignal while also masking the
effects of Rayleigh fading. Briefly, ~he
instantaneous amplitude of a received radio
~requency signal over a sequence of discrete time
intervals is samplad. The sampled signal levels
which do not axceed at least one of the signal
~L309143
- 7 - 45MR 511
lavel sampled immediately prior thereto in the
sampling sequence and the signal level samplad
immediately subsequent thereto in the sequence are
discarded. An average is calculated in response
to the levels not discarded. This average value
is outputted as an indicator of the average
received signal strength (RSSI).
The present invention produces a
received signal strength indicator which is
sensitive to sudden changes in average signal
level and can be dete.rmined rapidly. Moreover,
the number of measurements performed by the
present invention may be made small without
risking inaccuracy due to the effects of Rayleigh
fading on reaeived signal amplitude. Hence, a
small number of samples can be used to provide an
accurate average xeceived signal strength value
which is sensitive to rapid changes in the average
received signal strength and yet is largely
unaffected by deep fades.
DETAILED DESCRIPTION OF
PRESENTLY PREFERRED EMBODIMENTS
FIGURE 2 is a schematic block diagram
of a presently preferred exemplary received
signal strength indica~or system 100 in
accordance with the present invention. The
system 100 includes a radio receiver 102, an
analog-to-digital (A/D) converter 104, a memory
106, and a digital signal processor 108. System
100 may further include a digital output device
110, a digital-to-analog (D/A) converter 112, and
an analog output device 114. In the preferred
embodiment, A/D converter 104, D/A converter 112,
digital output device 110, processor 108, memory 106
, .
- ~L309143
--8--
and receiver 102 communicate with one another via a
conventional data bus 116.
Radio receiver 102 is a conventional radio
receiving device capable of rapidly tuning to any
frequency within its operating xange and having an
analog output which provides a measure of the
instantaneous amplitude of radio fre~uency signals
(of a desired frequency) received by an antenna 103
connected to the receiver. In the literature, such - -
an instantaneous signal strength output has beencalled the "Received Signal Strength Indicator"
(abbreviated RSSI). In the preferred embodiment,
receiver 102 comprises a General Electric Cellular
Station Radio Channel IJnit as described in the GE
15 publication designated LBI 31322; or General Electric
Cellular Mobile Radio described in the GE publication
designated LBI 31355.
Receiver 102 produces an analog electrical
output signal "RSSI", the level of which is a
function (e.g., logarithm) of the instantaneous
amplitude of radio freguency (RF) signals at the
frequency to which the receiver is tuned. The analog
output of receiver 102 is applied to the input of
conventional A/D converter 104, which converts the
analog output to a digital signal. The digital
output of ~/D converter 104 is applied to an I/0 port
of processor 108 via bus 116. Processor 108 may also
produce digital signals which are applied to a
digital input of receiver 102 via bus 116 to control
(in a conventional manner) the freguency to which
receiver 102 is tuned.
Processor 108 may be any conventional
microprocessor, and preferably includes a central
g ~L309143
processing unit, internal registers and counters, a
clock oscillator, and the like, all of which are
well-known. Processor 102 stores digital signals in
memory 106, and also reads digital signals from the
memory. The processor produces one or more output
signals which may be transmitted to central
controller 16 via digital output device 110 and/or
converted to analog signals (via D/A converter 112)
for graphical display on a chart recorder or other
display device 114 (e.g., in order to plot signal
strength for propagation studies). Processor 108
performs a series of predetermined steps under the
control of program instructions stored in a read only
memory (not shown) which may be internal or external
lS to the proces~or.
Receiver 102 monitors a frequency which may
be determined by processor 108 and continuously
~roduces an analog RSSI signal. A/D converter 104
converts this RSSI signal to a digital value. A/D
converter 104 in the preferred embodiment is of the
type which periodically samples the analog RSSI
signal and updates the digital value available at its
output in response to the then current received RF
signal amplitude. In the preferred embodiment,
processor 108 periodically reads (samples) the
digital value output by A/D converter 104 and stores
this value in an internal register NEW VALUE (NV).
Thus, the contents of the register NEW VALUE is the
current (i.e., most recently sampled) value
representing RSSI.
Processor 108 may ~ample the output of A/D
converter 104 at any desired sampling rate slow
enough to mask the effects of Rayleigh fading ~he
.
-lo- ~3~ 3
fixed sampling rate of the preferred embodiment i5
chosen to be on the order of the rate of occurrence
of Rayleigh fading of the received radio frequency
signal). When the processor samples the current RSSI
value, it first stores the contents of internal
register NEW VALUE into another internal register
called OLD VALUE (OV). The internal register OLD
VALUE thus contains a previously measured value of
RSSI tand, in the preferred embodiment, the RSSI
value sampled just prior to the most recently sampled
value). As will be understood, storage locations in
memory 106 could be used instead of internal
registers if desired.
Memory 106 stores an array 120 of values
A(l), A(2), ..., A(i), ..., A(N) (where N is a
positive integer). This array 120 is stored in
memory 106 separately from the OLD VALUE and NEW
VALUE registers internal to processor 108. A counter
("COUNTER", or "C") internal to processor 108 is used
to address (index) elements of array 120.
FIGURE 3 is a flowchart of the steps
performed by preferred embodiment 100. Processor 108
periodically samples the digital output of A/D
converter 104 and stores this digital value in
internal register NEW VALUE (block 204). However,
before the A/D converter output is written into
register NEW VALUE, the previous contents of the
register are (~r already have been) stored into
internal register OLD VALUE tblock 212) (thus
overwriting the previous contents of the OLD VALUE
register). At any given time, internal register NEW
VALUE contains the digital value representing the
RSSI most recently sampled from the output of A/D
1309~L43
converter 104 (e.g., at time ti), and internal
register OLD VALUE contains a digital value
represe~ting "next most current" RSSI (that is, RSSI
is sampled one sample time ti 1 prior to the most
recent sample time, or at ti ~ ~ where I is a fixed
sample period).
Processor 108 next compares the contents of
the OLD VALUE register with the contents of the NEW
VALUE register (block 2063 (for example, by executing
a conventional "compare" microinstruction which
generates a logical value indicating which register
contains the largest value). The larger of the two
values is stored in the location of array 120 pointed
to by the internal COUNTER of processox 108 (blocks
20B, 210), and the contents of COUNTER is then
incremented (i.e., increased by 1) (block 216).
The process~described abov~ continues
periodically until the value of COUNTER has reached N
(i.e., until all N array elements A(l) - A(N) contain
a value of RSSI, as tested for by block 214). That
is, processor 108 reads a series of signal amplitude
samples S1-SN+l representing instantaneous received
~F signal amplitudes at a corresponding sequence of
sample times tl-tN+l and stores N values, selected by
comparison to adjacent values, into memory 106.
Processor 108 then computes an average of the values
stored in array 120 by performing the following
calculation:
N
~ A(i)
i=l
AVG =
N
, ~ . .
-12- ~09~3
The calculated value AVG represents RSSI over the
time period tl through tN~l. This AVG value may be
sent to central controller 16 of the cellular system
10 (via digital output device 110) to facilitate hand
off determinations (or could be printed by a printer
connected to the output of output device 110) and/or
may be output via output device 114 and D/A converter
112.
The value AVG calculated by processor 108
provides a fairly accurate estimate of the actual
average value of RF signal amplitude received over
the sampling interval t1-tN+l. The number of samples
may be relatively small without risking that the
value AVG will be unduly influenced by a value
sampled during a fade. The close correspondence
between the value AVG determined in accordance with
the present invention and the actual average value of
received signal strength may be best understood by
referring to FIGURES 5tA) and 5(B).
FIGURE 5(A) is a graphical illustration of
an exemplary received RF signal plotted versus time.
The exemplary received RF signal strength shown in
FIGURE 5(A) is at or near its true average value most
of the time, although Rayleigh fading causes it to
fall substantially below this average value for short
periods.
Assume that the preferred embodiment lO0
produces a value AVG based upon five samples (i.e.,
samples taken at times tl, t2, t3, t4, and t5).
Thus, N = 4 (since N I 1 samples are reguired to
obtain the N values stored in array 120 in the
preferred embodiment).
, . . .
-13- ~3~9~43
At time tl, the sample value Sl of the
received sign~l first amplitude peak level is stored
into internal register OLD VALUE (see block 202 of
FIGUR~ 3). Subsequently, at time t2, the amplitude
S2 of the received signal is stored in register NEW
VALUE (block 204). Because the RSSI at time tl was
greater than the RSSI at time t2, the contents of
register OLD VALU~ is stored into element A(1) of
array 120 (blocks 206, 210). The RSSI existing at
time t2 is then loaded from register NEW VALUE into
register OLD VALUE (block 212).
Next, a value S3 representing the
instantaneous RSSI existing at time t3 is loaded into
register NEW VALUE (block 204). Since the RSSI at
time t2 (the contents of OLD VALUE) is greater than
the RSSI at time t3 (NEW V~LUE), the t2 value is
loaded into element A(~) of array 120 (blocks 206,
210).
In a similar manner, the value S4
representing RSSI at time t4 is loaded into array
element A(3) (since S3~S4), and the value S5
representing RSSI at time t5 is loaded into array
element A(4) ~since S4<S5). The value AVG is
calculated (block 218) from the values Sl, S2, S4 and
S5 stored in array 120 (representing RSSI at times
tl, t2, t4 and t5, respectively).
The value S3 representing RSSI at time t3
is not used to compute the value AVG, since the two
values it has been compared with (the value of RSSI
at t2 and the value of RSSI at t4) are both larger
than it. In the preferred embodiment, an RSSI value
Si measured for a given sample time ti which is less
than the RSSI value Si 1 measured during the
-14- ~309~3
immediately preceding sample time ti 1 and is al~o
less than the RSSI value Si~l measured during the
immediately ~ubseguent sample time ti~l does not
affect the calculated average value AVG at all, and
is instead discarded. A sample occurring during a
Rayleigh fade ~e.g., at sample time t3) has little
chance of affecting the calculated average value AVG
since it mGst likely will be surrounded by samples
not occurring during such a fade (so long as the
sample period is chosen to be on the order of the
duration of ~ost Rayleigh fades).
It will be observed that the calculated
average value ~VG is less than the peak signal value
and is a closer approximation of the actual average
value of the received signal strength than is the
peak value.
FIGURE 5(B) is graphical illustration of
the received signal strength (versus time) of an
exemplary rapidly changing RF signal. Such a signal
might be received rom a mobile station entering the
"shadow" of a large building. For the signal shown
in FIGURE 5(B), preferred embodiment 100 stores the
value Sl representing RSSI at sample time tl in array
element A(l) (Sl>S2), the value S2 representing RSSI
at time t2 in array element A(2) (S2 > S3), the value
S4 representing RSSI at time t4 in array element A(3)
(S3 ~ S4), and the value S4 representing RSSI at time
t4 in array element A(4) (S4 > S5). The value
representing RSSI at time t4 is stored twice (into
both array elements A(3~ and A(4)) because it is
greater than the values representing RSSI at times t3
and t5.
-15- ~3~9~43
The peak value of the received amplitude of
the signal shown in FIGURE 5tB) is substantially
greater than the true average received signal
amplitude due to the rapidly changing (decreasing in
the example shown) characteristic of the signal. The
value AVG calculated in accordance with the present
invention nevertheless provides a relatively accurate
approximation of the true average value even based on
a relatively small number of samples (five in the
example described).
The calculated average value AVG is
sensitive to rapid changes in the average received
signal strength and yet is not unduly influenced by
deep fades of short duration. The number of samples
N+l upon which the average value AVG is based
determines the sensitivity of the calculated average
val~e AVG to rapid changes in average received signal
ætrength (e.g., a large value of N will make the
calculated average value AVG relatively insensitive
to rapid changes, while small N causes the value AVG
to be more sensitive to rapid changes).
To ensure an adeguate and consistent number
of samples in any fixed interval of time, the present
invention occasionally stores a value representing
~5 RSSI at a given sample time into two elements of
array 120. For the exemplary signal shown in FIGURE
5(B), the preferred embodiment stores the value
representing RSSI at sample time t4 twice, causing
the calculated average value AVG to be biased
slightly. This biasing is intentional a~d ~esirable
for the following reasons. If the value sampled at
time t4 were not used twice (e.g., if a value
representing RSSI at any given æample time could be
,, ~ .
-16- ~3~9~3
stored into only one element of array 120 at most),
the calculated average value AVG would be based upon
fewer measure~ values (assuming a fixed number of
samples are performed over a fixed interval of time).
A further sample at a time t6 later than time tS
would then be necessary to determine whether RSSI
measured at sample time t5 should be stored into
array 120 (as can be seen, the value representing
RSSI at time t4 is a better estimator of the average ~
received signal strength than the value measured at
time t3 or t5). The preferred embodiment 100 never
stores an RSSI value obtained from a given sample
into more than two elements of array 120.
A process which gathers extra samples
(e.g., at time t6j to make up for omitted samples
extends the time re~uired to obtain a result, and is
therefore.undesirable. It is possible to use sample
periods of variable length (e.~., to sample again at
t5 ~ ~t < t5 -I ~, where ~ is the nominal sample
period, i~ the RSSI value measured at time t5 is to
be discarded), but additional complexity would be
introduced which probably would not significantly
improve the results obtained.
Preferred embodiment 100 provides excellent
results when used, for example, in a cellular
locating receiver which is reguired to perform rapid
measurements on a number of different channels. In
such device, receiver 102 is directed by processor
108 to tune to each of a predetermined sequence of
channels, and to perform the steps shown in FIGURE 3
on each channel.
Sometimes, however, it is necessary to
perorm RSSI measurements repetitively while receiver
~3~91~3
-17-
102 is tuned to a single channel. For example, a
mobile station measuring propagation of a fixed
station probably should remain tuned to a single
frequency for the entire duration of the propagation
test. The preferred embodiment lQ0 can be modified
slightly to calculate a rolling average and to thus
output a calculated average value AVG for each and
every sample period if desired.
FIGURE 4 is a flowchart of steps which
provide a rolling average of RSSI in accordance with
the present invention. The contents of the COUNTER
internal to processor 108 is reset to 1 whenever its
contents are incremented to greater than N (blocks
224, 226). An accurate calculated average value AVG
is first available as soon as N ~ 1 samples have been
read by processor 108, and can be calculated
accurately at any time thereafter (e.g., on demand,
ater every sample time, or at any other convenient
interval) (blocks 222, 218). The sample siæe N~l may
be chosen to adjust the sensitivity of the
measurement to rapid changes, and samples occurring
during fades will not unduly influence the value of
AVG.
FIGURES 3(A) and 4(A) are additional,
modified flowcharts showing steps which perform the
same basic calculations as do the steps shown in
FIGURES 3 and 4.
The advantages and improvements the present
invention provides over the prior art techniques for
measuring RSSI should now be apparent. The present
invention provides better accuracy by eliminating
readings in deep ades and averaging the values that
are left. The technigue di closed in U.S. Patent No.
.
-18- ~09~43
4,549,311 to McLaughlin would merely select peak RSSI
value occurring at time tl for the exemplary signals
shown in both FIGURES 5(A) and 5(B). The average
value AVG determined in accordance with the present
invention is much closer to the true average signal
level for each of these exemplary signals.
Although a peak reading may be closer to
the true average than a reading taken during a deep
fade, the difference between peak received signal
strength and average received signal strength is
significant in many (if not most) cases. The peak
approximation is not accurate for received signals
which fluctuate in strength due to factors other than
Rayleigh fading. The present invention calculates a
value which is sensitive to these other factors as
well as more accurately estimating the average
received signal strength (while also excluding
measurements taken during deep ades), and therefore
provides a far more accurate indication of average
received signal strength.
While the present invention has been
described with what is presently considered to be the
most practical and preferred embodiments, it is to be
understood that the appended claims are not to be
limited to the disclosed embodiments, but on the
contrary9 are intended to cover all modifications,
variations and/or equivalent arrangements which
retain any of the novel features and advantages of
the invention. By way of non-limiting example,
analog techniques and structures could be substituted
for the various digital techniques and structures of
the preferred embodiment ~e.g., analog sample and
hold circuits may be us~d to store analog RSSI
' , :
'~ .
-19- ~3~9il~1L3
values, and operati~nal amplifier techniques may be
used to obtain a signal representing an average of a
plurality of analog signal levels). Processor 10~
can, of course, perform many other functions as well
as the steps shown in FIGURE 3. If desired, all
sampled RSSI values could be stored in memory 106 and
the analysis of the present invention could be
performed on the stored values (such analysis, not
being performed in real time, could begin at the end --
of the stored sequence and compare values sampledlater with those sampled earlier). Alternatively,
memory 106 could be eliminated and a running sum of
selected sampled values could be maintained.
Moreover, the signal processing steps of the present
invention can be implementing using hardware,
software, firmware or any combination of these.
.: -
'' '~
