Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
-1- 21~ ~
DI~ AUDIO ~EQUENCY TONE DETECTION :
S l~he pre~ t inventioD r~ilate~ gen~ly to ~he field of
commu~ication~ and par~ arb to to~0 de~o~.
Landline t~lep~ony we~ ~uperi~lsion to de~ect changes
in the ~v~tch-hool~ atate caused by t~lB telephoIle user. Mobile
telephone supervi~ion abo per~on~ 1;hi~ proce~s but must
~so ensure that adequate RF ~ignal streDg~ and interference
protection i8 maintai~ed. ThiB is accomplished by 1 h~
supervi~org audio tone (SAT), a conl;inuou~ out-of-band
modulated signal.
T~ree SAT 8ig~1e, are u~ed D3 the UDited State~
cellular system, AMPS. These SAT ~ignal~ are at 5970 Hz,
6000 Hz, and 6030 Hz. Onlg one of the~e ~equen~es iB
2 0 employed at a given cell site.
The SAT operate~ by the bile unit recei~ring the SAT
from the base station and transponding it back to clo~e the
loop. l~e base station 1001~B for the return of the ~pecific SAT it
~ent out. If another SAT i~ returned, the c~ll interpret6 thi6 a6
2 5 the call between the mobile and the cell being corrupted by
interference.
When a mobile receives a aig~l from the ba~e ~tation, it
detect~ whether SAT is pre~ent. Baset on this frequency, the
mobile generates its own SAT ~d tran~mits it bac~ to the base
3 0 station. l~ method requires t~e mobile to perform a larg~
number of million in~truclion~ per ~econd (MIPS) in order to
detect t~e received SAT and generate the t=itted SAT.
Consumers are demanding smaller cellulsr telephones
for greater porhbility. To reduce 1 he l~ize of the telephone~, the
3 5 number of p~tB in t}le ~lephone must be reducet. Thi~ can be
accomplished by performing many of the telephone'~
-; : . :;, - . ,
~ 2~1~a52
-2-
fimclions in a digit~l ~ignal proce~or (DSP3. Thi8, ill effect,
r~places a nu~er of integrated ~ts wit~ a ~le DSP
~hat pe~01~18 1~he same fi~clio~ e replaced IC~.
Replacing t~e present SAT detec~o~ and generation
5 circuit~ wi~ a DSP would ~ a ve~y M~S intensive
proce~s. Tbi~ would ta~e up time 1he D~P could bs used for
other ta~. Addi'donally, the rs~ived 8AT mu~t be adju~ted
for ~ di~erences cau~sd by the mobile'~ hard~vare before it ~ :
i~ tra~itted ba~lc to ~he base 8tatiOII. T~i8 would take
1 0 additiol~al 1ime away fiom the DS~. Tnere i~ a re~ulting need
for a simple proce~ to detect SAT and adju~t it~ gain.
The procee~ of t;he pre~ent ~ventio~ encompa~ a
method for automatically coI~trolling the gain of a t~a~mitted
modulation BigIlal. Thi8 control iB ba~ed on a received
demodulated ~ignal. The proce~s filter~ the meived signal to
produce a Sltered ~ignal. Thi8 filtered 8igllal i8 proces~ed by
2 0 an autocorrelation fi~nction to g~nerate a plurality of
autocorrelation values. A desired energy of the filtered signal
iB divided by at least one of the autooorrelation values to
produce a preliminary ga~ adju8tment ~ al. A filrther
gain iB derived from the prel;minar~ g~n adjustment Big
2 5 by tal~g the ~quare root of the prelimi~a~ gain adjusl;ment
Bignal. The derived gain i8 scaled in response to lhe gain of
the ~nod~ator that modulate~ 1 he trulsmitted signal. This
derived gain i~ filtered to produce a filtered gaiD signal. The
transmitted signal is generated in response to the filtered gain
3 0 signal and ~e filtered ~ignal.
2 ~ 5 2
-3-
ElhOWB a blo~lc diagram of ~he plOCl~ElE of the
present in~en~on.
S ~IG. 2 ElhOW8 a block diagram of the &~T detection
proCe8a of the pr~sent inventio~
1 0 The prlU:e88 of ~he pre~ent inventio~ enable~ a ~;AT - -
detection and automatic ~ control (AGC) p~oce~
~orated into a D~P ~vithout r~quirine a large amount of
p~B8~g 1ime. ThiB prOCe811 iB illustrated m 1~e block
diagram of EIG. 1.
T'ne proo~ss begins by 1~e received ~ignal being
demodulated (102). T'nis de~nodulation prOCeBI~ produces a
~al who~e rm8 value is affiected by both the rece~ed BigII~I
and the gun of the demodulator. Tni~ af~ec~ ia removet by
multipl~g (103) the demodulated ~ignal by the value N. N is
2 0 chosen BO that for a known received sigDal, the signal at the
input to the bandpa~s filter (104) is Icnown; i.e., N is adjusted
for each radiotelepho~e to remo~e the effects of demodulator
gain varianc~ f~om radiot~lephone to radiotelephone.
To illustrab this adjustment, one embodiment of the
2 5 pre~ent invention might ha~re a received Bignal with an
appronmate 2.0 kHz deviation at a 6.0 kHz rate. The actual
received ~ignal may be other than e~ y 2.0 kHz de~Tiation,
but the trsnsmitted signal needs to be e~ y, or substantially
close to a 2.0 k~z de~iatio~ independent of the esact received
3 0 signal deviation. The tran~ponded rece*ed signal, therefore,
needs to have its gain adju~ted as the demodulated rece*et
signal varies.
For a received signal with a 2.0 kHz devia~o~ at a 6.0
kHz rate, a demodulated signal, a(t) is producet. This 9ignal
3 5 i8 1hen sampled and ga~ adju~ted to produce a diacrete
signal, s(n), with an rms v~lue D. ~(n) is then filtered by a
'.'`"'`"', '", ,, ~
' "' '~
2 ~ 2
4 -
narrowband bandpa~ filter tlO4) ~harac~rizsd by tran~fer
fmlction h(n) wh4~e output, ytn) ha~ value E. The BPF
(104) remOVeB e~h and noi~e ~o 1~at ytn) i~ a fairly c3o~e
repre~entation of SAT.
S y(n) i~ input to an autocorr~lation fi~ction calculator
(106)that generates outpub r~(O), rs(1), and r~(2). Th~se ou~put
ralues are the e~rg~r of each sample of the sampled Bignal,
~(n). r~(O) for an input sine ~vave of nns value E ha~ value 132.
The radiotelephone's t~tbr al~o has a variance
iiom radiot~lephone to radiote~ephone ~br effective tran~it
deviation. To produoe a }nown de~atioll, the modulation level
must be adjusted to compensate for t~i~ ~uiance. T'ni~ is done
by adjusting the value of the constant L for ea~ individual
radiotelephone .
The proeess of the present ill~ention performs SAT
modulation gain adjustment by u~g the E2 value to derive a
gain adjus~ent for the output ~ignal pnor to its pas6age to - -; -~
the tran~itter. Tbis adjw~ent is performed by t}le proces~
ofthe present invention by sc~ling the signal, ~(n) to produce a
2 0 signal of amplitute G. An input sinusoidal signal of rm~
~alue G is required at the input of the t itter to produce a
2.0 kHz denated sigDal at the tran~itter output.
G is determined by the follo~ving equation~
2 5 G = L
A2 :.
where the values for - 2Dl' and L are cllosen ~o that for an
autocorrelation output of E2, a ~ignal of rms value G iB
generated if y(n) is at the correct f~equen~y and v.1ithin an
3 0 amplitud~ windo~r, as determined by ~he SAT detector.
The output of the abo~e op~ration is input to a low pallB
filbr (LPF) (106). T~e LPF (106) reduoe~ 1 he variance of the
gun ad,justment siFnal to keep the tran~mitted aigllal from
e~periencing large amplitude cha~ges.
~ . ~ . , ... . . . . .. ,, . .. ~, .. .... ... ... ... ..... . ..... ..... .... . . . .. . .. .
.' ., ' ' " ' ' ' ~ . . ' . , . ' :
2~8.S5~
T'ne output of th~ LPF (106) is one ~ut to a ~wit~hing
opera~n (101). If S~T i~ det~d b~r th0 SAT debctor (110), the
ou~ut ofthe LPF (106) i~ ub0equeQt operation~. If
8AT i~ t detected, the gain of the traDsmitted sig~ et to
5 zero.
Tn~ SAT detector (110) of the pr~t inv~on
detern~ines the pre~ence of t~ ~T ~ignal by the amplitude
and i~uenc~ ~eri~ics of the received Bignal. To
aooompli~h thi~, ~e proce~s i~lustrated ~ ~IG. a iB u~ed.
Tkis proceas beg~ns by the autocorre~ation vs~ue~ ~om
1 he autocorrel~tion calculator being ~put to the SAT detector.
Using the Le~insoll-Durbin ~ve proce~s, direct form
~efficie~ts are generated from 1~ autoGorrelation ~alues.
The pole loca~ of the autocorrelation value~ are then
1 5 ge~lerated by ~putting the ~:oefficients into a quadratic
equation. l~he f~equency of 1he receqved signal i~
determined bg these pole location~. T'ne ~mplitude
~haracteristic is terived na r,(0) and and compared ~ith a
possibly variable thre~hold. If E2, as mea~ured by r~(0) is
2 0 abo~e the threshold, the received SAT is determined to have
sufficient amplitude to be detected, if 1 he f~equency, as
determined by pole locations, i8 acceptable.
The SAT frequency detection process can be represented
mathematically as follows. The z - traDsform of a ~ine wave is:5
sin(~ => z2 2zco~ T) + 1
where: ~ = 2~o (fo S z~ d T = sampling rate.
Tbi8 CaD alBO bewritten a~
~ :
. ( ) z~ DT)
1- 2z ~(~DT)~Z'2
The equation for a second order linear tim~-invariant
discrete-time ~ystem iB:
. .
~ -6- 2118~
H(z)' aO al~l ab~2
Equating ~e coefficienb of 1ihe ~ fimction wit~ 1 hat of
5 ~I(z) give~:
bo=O aO=l
q = 2cos(a~
b2 = o a2 = -l
''
~ince the ~equenc~ o 1 he rec~ved BigDal iB wanted, the
~ of a ~ oidal ~al can be esplit:i1 Iy repre~ented :
by 1he de~ominator of H(z). The numerator of H(z) ~res oDly
gain infonnatio~L U~ t~e ~a~c equation to find the :
pole locatioDs in the z-doma-n give8: ~
a2(1) ~ (1)+4a2(0)ag(2) ~ ~:
z=
The frequency of the recei~ed ~inusoid~l ~ignal i6: :~
2 0
o=tan~ (1)+2~1a)2()a2( )3~
The denominator coefficients of the ~econd-order linear
time-invanant di~crete-time ~y~ be ea~ily detennined ~ -
2 5 by u~ he Lenn~on-Durbin recur~ive process. The
Lenn~on-Durbin recur~ive proce~ for detormining 1 he direct
form coefficient~ iB a~ foillow~, where r~m) i~ ~he unbia6ed
edimate of the true autocorr~ation.
N-Y-l
3 r~S(m) = N - M n~O n + ~ X~[n]
where: X(n) = discrete data ~ample,
-. ,;~. ;. . i .. .~ . .. . .. .;. . . . .. .. . .
~~ 7 2~1 8~52
. :
N = anal~ 1
Now r,stO), r~(l), alld r"(2), the autooorr~ation ~alue~
fra~ the autocorrelation calculator, are wed to calculat~ the
~ form coefficients.
~0)= 1
a2(1) = al(l) + r2 al(l)
a,(2) =r2
1 0 where a~(l) = r~ = r,~O)
l~s(2) ~ al(l) r~(l)] r~
r2= ~(U-~tO) :
r~2~r~,(0) . r~(1) 1 -~
r2 r~"(l) - r~(O)
1 5 ' ~, :
a2(o) S 1
a2(1) = r~(V - [(r (1) r~((O3)(r~(l)) + r (0))3
2 r~(2)r~u(0) - ~(1)
a2( ) = ~,~(1) - ~()
a2(1)=r~(l ~( ) ~( )]
If the frequenc~ of the ~ignal iB determ~et to be eit her
~970 Hz i lO Hz, 6000 Hz i lO Hz, or 60~0 Hz i lO Hz, &~T i~
2 5 present in the receiYed Etignal. If SAT is pre~ent, ~he SAT :
detector (110) enable~ t~e s~itchi~ nction (101) to allow the
gain adjusted ~ignal t~rough If SAT is not present, the SAT
detector (110) enable~ 1he swits l~g function (101) to output a
mute control (120) ~ignal.
' ' ;,':
.,