CN1128529C - Multi-channel 64Kbps squelch compression method for packet switch network - Google Patents

Abstract

The present invention discloses a multi-channel 64Kbps voice muting and compression method which is suitable for packet switching network. By the method, muting compression is carried out for data which can be compressed, such as voice or voice band data modulated by a DODEM at the sending end, and the data is transmitted; time slots without signals are not processed; service which presses time slots but is not allowed to be compressed, such as signaling and ISDN data can be transmitted by a transparent method, and thus, the bandwidth utilization rate of a transmission channel can be increased. Meanwhile, because the present invention uses the state analysis of time slots for the static compression of data, the present invention is simple and easy to realize.

Description

Be applicable to the multichannel 64Kbps speech silence compression method of packet network

Technical field

The present invention relates to field of telecommunications, be specifically related in packet network, transmit in the field of telecommunications method of multi-path digital voice service.

Background technology

At present, the professional method commonly used of transmission multi-path voice is that 64KbpsPCM (pulse code modulation) digital voice business is directly transmitted in packet network in packet network, speech transmissions Bell Laboratory technical journal in document the atm network in 1998, the 10-12 monthly magazine, 176-191 page or leaf (Mahdi S.Chambers, Harbans Kaur, Terry G.Lyons, andBrian P.Murphy, Voice over ATM, Bell Labs Technical Journal, October-December 1998, P.176-191) mention can adopt the silence detection technology that voice are compressed after, send into ATM (Synchronous Transfer Mode) Network Transmission, how not realize but relate in atm network.In voice communication process, normally side speech, the obedient and sounding not of the opposing party.In addition, speech one side also has pause, the time of the actual speech of communicating pair, i.e. active stage, in whole communication process, only account for 30%-40%, and when therefore 64Kbps PCM digital voice business directly being transmitted in packet network, the bandwidth availability ratio of network is very low.For many input digit telephone lines, if use DSI (digital speech interpolation) technology, only when sound, pass the signal to the other side, will not transmit during sounding, just can improve bandwidth availability ratio, increase the message capacity of system.For the voice band data service of adopting MODEM (modulator-demodulator) modulation, as fax, a side sends data often, and the opposing party only sends out some confirmation signals, and a large amount of free time is arranged, and can adopt the DSI technology to improve bandwidth availability ratio equally.

Because voice signal is in active stage or nonmobile phase is at random, the speed of signal no longer is a steady state value after the silence compression, for guaranteeing the constant of output speed, traditional silence compression equipment must carry out dynamic load control, and dynamic load control realizes very complexity, make and apparatus expensive limited its application.

Summary of the invention

At above-mentioned prior art problems, the purpose of this invention is to provide a kind of multichannel 64Kbps speech silence compression method that is applicable to packet network that simply is easy to realize.

For achieving the above object, the technical solution used in the present invention is: a kind of multichannel 64Kbps speech silence compression method that is applicable to packet network, and this method comprises data transmission procedure and DRP data reception process, wherein:

Data transmission procedure may further comprise the steps:

(A) initialization is made as state of activation with the status register of all time slots, activate counter, quiet counter zero setting continuously continuously, and set handling duration T presets quiet constant K, M, N, activates constant P and signal sample constant S;

(B) read timeslot number since 0;

(C) time slot that above-mentioned steps (B) is read is judged, if last time slot disposes, changes step (D) and continues operation, continues operation otherwise change step (E);

(D) wait for end, change step (B) then and continue operation to epicycle handling duration T;

(E) extract the compression sign of these data that read timeslot number;

(F) according to above-mentioned steps (E) but the compression sign judgment data that obtains silence compression whether as not allowing compression, change step (G) and continue operation, continue operation otherwise change step (H);

(G) add timeslot number before data, the data of this time slot of transparent transmission are changeed step (B) then and are continued operation;

(H) the state decision according to current these time slot data abandons or these time slot data of transparent transmission, changes step (B) then and continues operation; DRP data reception process may further comprise the steps:

(a) according to timeslot number the data that receive are packed in the corresponding time slot of the corresponding frame of receiving terminal;

(b) fill the idle time slot of receiving terminal with white noise.

Above described state decision according to current these time slot data abandon or these time slot data of transparent transmission are carried out according to following steps:

(A1) judge whether L signal sample in this handling duration T has S greater than thresholding,, forward step (A2) to and continue following operation, continue following operation otherwise change step (A3) if having;

(A2) carry out follow-up judgement and quiet processing;

(A3) the value R of read states register is a state of activation as R, changes step (A4) and continues operation, as R is mute state, changes step (A7) and continues operation, as the R mute state that is as the criterion, change step (A12) and continue operation,, change step (A13) and continue operation as the R state of activation that is as the criterion;

(A4) quiet counter adds 1 continuously;

(A5) do you whether judge quiet number of times=K? if be not equal to K, change step (A20) and continue operation, continue operation otherwise change step (A6);

(A6) the configuration state register mute state that is as the criterion is changeed step (A20) then and is continued operation;

(A7) quiet counter adds 1 continuously;

(A8) activate counter puts 0 continuously;

(A9) whether judge quiet number of times 〉=M,, change step (A10) and continue operation, continue operation otherwise change step (A19) as greater than M;

(A10) the configuration state register is a mute state;

(A11) quiet counter puts 0 continuously, changes step (A20) then and continues operation;

(A12) abandon to return after the data and handle next time slot;

(A13) quiet counter adds 1 continuously;

(A14) judge whether quiet number of times equals=N, as equating that changeing step (A16) continues operation, continues operation otherwise change step (A15);

(A15) activate counter puts 0 continuously, changes step (A20) then and continues operation;

(A16) the configuration state register is a mute state;

(A17) quiet counter puts 0 continuously;

(A18) activate counter puts 0 continuously, changes step (A20) then and continues operation;

(A19) return next time slot of processing after transmission timeslot number, noise average power and the data;

(A20) transmit timeslot number and data, return and handle next time slot.

Above describedly carry out follow-up judgement and quiet processing is carried out according to following steps:

(C1) the value R of read states register is a state of activation as R, changes step (C16) and continues operation, as R is mute state, changes step (C2) and continues operation, as the R mute state that is as the criterion, change step (C8) and continue operation,, change step (C10) and continue operation as the R state of activation that is as the criterion;

(C2) activate counter adds 1 continuously;

(C3) judge that the activation number of times whether more than or equal to P, is not more than P if activate number of times, change step (C4) and continue to operate, continue operation otherwise change step (C5);

(C4) quiet counter adds 1 continuously, changes step (C15) then and continues operation;

(C5) the configuration state register is for activating;

(C6) activate counter puts 0 continuously;

(C7) quiet counter puts 0 continuously, changes step (C16) then and continues operation;

(C8) the configuration state register is for activating;

(C9) activate counter adds 1 continuously, changes step (C16) then and continues operation;

(C10) activate counter adds 1 continuously;

(C11) judge whether equal P, if equal, the step of walking around (C12) continues operation if activating number of times, otherwise step (C14) continues operation;

(C12) the configuration state register is for activating;

(C13) activate counter puts 0 continuously;

(C14) quiet counter puts 0 continuously;

(C15) return next time slot of processing after transmission timeslot number, noise average power and the data;

(C16) return next time slot of processing after transmission timeslot number and the data.

The technical scheme that is adopted by the invention described above because the present invention adopts at transmitting terminal, to the data that can compress, is carried out the silence compression transmission as speech or through the voice band data of DODEM modulation as can be seen, and the time slot of no signal is not processed; To taking the method for forbidding the business compressed such as signaling and the transmission of ISDN (integrated services digital network) data transparency in the time slot, can improve the bandwidth availability ratio of transmission channel, simultaneously because the present invention is used for the Static Compression of data with the time slot state analysis, method is simple and be easy to realize.

Description of drawings

Fig. 1 is a data transmission procedure main flow chart of the present invention;

Fig. 2 is current time slots flow chart of data processing figure of the present invention;

Fig. 3 is the quiet process charts of data of the present invention;

Fig. 4 is a hardware adaptations environment schematic diagram of the present invention;

Fig. 5 uses silence compression equipment schematic of the present invention;

Fig. 6 is the transition diagram that takies between the time slot operating state;

Fig. 7 is the hardware environment figure of the embodiment of the invention;

Fig. 8 is AAL2 CPS adaptation procedure figure;

Fig. 9 is AAL2 CPS packets headers and CPS_PDU domain definition format figure;

The cluster speech formed speed R behind the cell and the graph of relation of traffic carrying capacity a when Figure 10 was different γ;

The graph of a relation of η and γ when Figure 11 is different a;

Figure 12 is the probability graph that the cell bit rate surpasses bandwidth B.

Embodiment

Below in conjunction with drawings and Examples technical scheme of the present invention and advantage are further described.

At first with reference to figure 4, when multi-path voice signal access package switching network transmits, at first carry out quiet processing through silence compression equipment, signal after the processing is packaged into the packet format that needs by the packet switch connecting system and sends into packet network, when signal is exported by packet network, at first unpack, have the voice service restore circuit to form the PCM frame format of multi-path voice signal demand again by the packet switch connecting system.

At transmitting terminal, silence compression equipment to need to obtain the business of transparent transmission by signaling, as ISDN (integrated services digital network) data and signaling data, and the position in the cluster speech channel, i.e. timeslot number does not process the time slot of no signal; And forbid the professional transparent transmission compressed in the time slot to taking; Carry out silence compression to speech or through the voice band data of DODEM modulation.With reference to figure 5, silence compression equipment is input as the PCM cluster frame signal from the black phone switching network, divides each speech channel by time slot.Be output as through the voice signal after the silence compression, L sample value of every road speech carried the set of time-slot sign, is generally 10 bit time slots number and 3 bit noise average powers.

Silence compression equipment only carries out silence detection to the operating state that current each road takies time slot, and idle time slot is left intact.With reference to figure 2, each road signal that the generalized shuffle network among Fig. 2 will take time slot rearranges, and set handling duration T puts the data of each road T in second together, adds that two byte timeslot numbers put into buffer.Timeslot number extraction circuit is that the voice signal insertion time slot after the silence compression is prepared, average power signal counting circuit and noise average power counting circuit are used for self adaptation and adjust the threshold level of silence detection, noise average power inserts in the signal simultaneously, pass to receiving terminal, provide the power foundation for receiving terminal carries out the continuous white noise insertion to quiet time slot.The compression control circuit judges that according to the signaling information of cluster frame the signal of current time slots could compress, the work of control silence compression circuit.The operation principle of silence compression circuit is: whether the signal of at first judging current time slots transmission is the data that can not carry out data compression, if, transparent transmission then.To carrying out the time slot that takies of data compression, its operating state is divided into activation, accurate activation, quiet and accurate quiet one of four states.The purpose of silence detection is exactly that correct judgement respectively takies the operating state of time slot in current handling duration T, is sent to the purpose module pellucidly with the data that determine current handling duration correspondence and still is used as quiet data and abandons.With reference to figure 6, the figure shows the transfer relationship that takies between the time slot operating state.Silence detection regular as follows, taking time slot with i number is example: at first compare with the adaptive threshold level the 6 byte digital signals that take time slot in the cluster voice signal for i number are byte-by-byte, the byte that digital signal level is higher than the adaptive threshold level is the state of activation byte, otherwise is the mute state byte; Again the comprehensive state that takies 6 bytes of time slot for i number is carried out accurate majority decision, if the state of activation byte number, shows then that the i time slot detects once more than or equal to the mute state byte number and activate, otherwise it is quiet to show that the i time slot detects once.

At transmitting terminal, silence compression equipment to need to obtain the position of business in the cluster speech channel of transparent transmission by signaling, i.e. timeslot number does not process the time slot of no signal; And forbid the professional transparent transmission compressed in the time slot to taking; Carry out quiet processing to speech or through the voice band data of DODEM modulation.With reference to figure 3.Silence compression equipment is once adjudicated the voice signal of L byte, and (O≤I＜when L) individual sample value is higher than detection threshold, is judged as this road signal and occurs once activating, otherwise be that appearance is once quiet more than I is arranged in L the sample value.I/L and constant K have determined the sensitivity of Voice Activity Detector, and K is more little, I/L is more little, and sensitivity is high more, but antijamming capability is poor more.

Silence compression equipment is divided into four kinds with the state of each road signal: state of activation, accurate state of activation, mute state and accurate mute state.In state of activation, transmit this road speech shared timeslot number and L byte data in the cluster speech channel pellucidly, the average power of while signal calculated, and this road speech carried out silence detection, quiet as if being consecutively detected K time, think that then this road signal does not have sound to need transmission, transfer to accurate mute state by state of activation, otherwise it is constant to keep state of activation.When transferring to accurate mute state, though this speech channel is in nonmobile phase,, still to continue this road signal regular hour that transmission comprises timeslot number in order to keep the naturalness of speech, this time is the suspension time, as 30ms.Hang the time timing so start the suspension time timer, proceed silence detection simultaneously.If be consecutively detected K activation, then return state of activation, if the timer meter does not detect K time continuous activation to suspension during the time, then transfer to mute state.In addition, can calculate the background noise average power of this speech channel this moment, and according to this noise average power with at the average power signal that state of activation is calculated the thresholding of silence detection is dynamically adjusted, to reach the optimum detection effect.Then,, the background noise average power is represented with 3 Bit datas, be sent to receiving terminal according to the ITU-TG.763 recommended standard.In mute state, abandon the L byte data of this speech channel, carry out silence detection simultaneously.Activate as long as detect once, just this time slot transfers accurate state of activation to by mute state, otherwise it is constant to keep mute state.This measure is in order to reduce the cut sound of quiet processing to signal, and promptly voice signal is from noiseless when forwarding sonance to, the one piece of data the when speech that needs detection time and lose owing to silence compression equipment is initial.In accurate state of activation, this road signal (containing timeslot number) of transparent transmission is proceeded silence detection simultaneously.If be consecutively detected K time quiet, then return to mute state, as if being consecutively detected K activation, then transfer state of activation to.As the above analysis, when certain speech channel was in mute state, data were dropped.When other states, with L byte voice signal and [log ₂M] bit (M road cluster speech, [log ₂M] be more than or equal to log ₂The smallest positive integral of M) timeslot number, 3 bit background noise average powers are sent to receiving terminal together.

G.763, G.712 and V.2 determine the initial examination and measurement threshold value of each road signal about the average power signal on the standard telephone line and noise average power according to ITU-T suggestion.But signal on the actual track and noise average power not necessarily satisfy above-mentioned condition, and time to time change, so silence compression equipment is according to the average power and the Background Noise Power of each the road signal that calculates, dynamically adjust detection threshold, with acquisition optimum detection effect, and the speed of thresholding adjustment is controllable.

At receiving terminal, the data that the voice service restore circuit must transmit atm network revert to the cluster frame signal.To being in the speech channel of non-mute state,, in the corresponding time slot with byte-by-byte in regular turn receiving terminal L the PCM cluster frame of packing into of corresponding L byte data, 3 Bit datas that carry the background noise average power are stored in the noise power memory according to the timeslot number that transmits.To being in the speech channel of mute state, send without any data, at first from the noise power memory, take out the background noise average power that transmitting terminal transmits in accurate mute state, form the continuous white noise of same average power, add the corresponding time slot that is charged to this road speech in the cluster frame.

Concrete implementation step of the present invention is as follows:

At first the implementation step of data transmission procedure is as follows:

With reference to figure 1.In step (1), carry out system initialization, the status register of all time slots is made as state of activation, activate counter, quiet counter zero setting continuously continuously, set handling duration T presets quiet constant K, M, N, activates constant P and signal sample constant S; In step (2), read timeslot number since 0; In step (3) time slot that above-mentioned steps (2) reads is judged,, changeed step (4) and finish the epicycle processing, continue operation otherwise change step (5) if last time slot disposes; Wait for end in step (4), change step (2) then and continue wheel operation down to epicycle handling duration T; Extract the compression sign of these data that read timeslot number in step (5); Step (6) according to above-mentioned steps (5) but the compression sign judgment data that obtains silence compression whether as not allowing compression, change step (7) and continue operation, continue operation otherwise change step (8); In step (7), before data, add timeslot number, the data of this time slot of transparent transmission are changeed step (2) then and are continued operation; In step (8), decision abandons or these time slot data of transparent transmission according to the state of current these time slot data, changes step (2) then and continues operation.

DRP data reception process may further comprise the steps:

At first the data that receive are packed in the corresponding time slot of the corresponding frame of receiving terminal according to timeslot number; Fill the idle time slot of receiving terminal then with white noise.

Above described state decision according to current these time slot data abandon or these time slot data of transparent transmission are carried out according to following steps:

With reference to figure 2.At step (A1), judge whether L signal sample in this handling duration T has S greater than thresholding, if having, forward step (A2) to and continue following operation, continue following operation otherwise change step (A3); At step (A2), carry out follow-up judgement and quiet processing; At step (A3), the value R of read states register is a state of activation as R, changes step (A4) and continues operation, as R is mute state, changes step (A7) and continues operation, as the R mute state that is as the criterion, change step (A12) and continue operation,, change step (A13) and continue operation as the R state of activation that is as the criterion; At step (A4), quiet counter adds 1 continuously; Do you at step (A5), whether judge quiet number of times=K? if be not equal to K, change step (A20) and continue operation, continue operation otherwise change step (A6); At step (A6), the configuration state register mute state that is as the criterion is changeed step (A20) then and is continued operation; At step (A7), quiet counter adds 1 continuously; At step (A8), activate counter puts 0 continuously; At step (A9), whether judge quiet number of times 〉=M, as greater than M, change step (A10) and continue operation, continue operation otherwise change step (A19); At step (A10), the configuration state register is a mute state; At step (A11), quiet counter puts 0 continuously, changes step (A20) then and continues operation; At step (A12), abandon to return after the data and handle next time slot; At step (A13), quiet counter adds 1 continuously; At step (A14), judge whether quiet number of times equals=N, as equating that changeing step (A16) continues operation, continues operation otherwise change step (A15); At step (A15), activate counter puts 0 continuously, changes step (A20) then and continues operation; At step (A16), the configuration state register is a mute state; At step (A17), quiet counter puts 0 continuously; At step (A18), activate counter puts 0 continuously, changes step (A20) then and continues operation; At step (A19), return after transmission timeslot number, noise average power and the data and handle next time slot; At step (A20), transmit timeslot number and data, return and handle next time slot.

Above describedly carry out follow-up judgement and quiet processing is carried out according to following steps:

With reference to figure 3.At step (C1), the value R of read states register is a state of activation as R, changes step (C16) and continues operation, as R is mute state, changes step (C2) and continues operation, as the R mute state that is as the criterion, change step (C8) and continue operation,, change step (C10) and continue operation as the R state of activation that is as the criterion; At step (C2), activate counter adds 1 continuously; At step (C3), judge and whether activate number of times more than or equal to P, be not more than P if activate number of times, change step (C4) and continue operation, continue operation otherwise change step (C5); At step (C4), quiet counter adds 1 continuously, changes step (C15) then and continues operation; At step (C5), the configuration state register is for activating; At step (C6), activate counter puts 0 continuously; At step (C7), quiet counter puts 0 continuously, changes step (C16) then and continues operation; At step (C8), the configuration state register is for activating; At step (C9), activate counter adds 1 continuously, changes step (C16) then and continues operation; At step (C10), activate counter adds 1 continuously; At step (C11), judge whether equal P, if equal, the step of walking around (C12) continues operation if activating number of times, otherwise step (C14) continues operation; At step (C12), the configuration state register is for activating; At step (C13), activate counter puts 0 continuously; At step (C14), quiet counter puts 0 continuously; At step (C15), return after transmission timeslot number, noise average power and the data and handle next time slot; At step (C16), return after transmission timeslot number and the data and handle next time slot.

Above described handling duration T, quiet constant K, M, N activate constant P and signal sample constant S and determine by emulation according to system condition.

Below by the method that realizes multichannel 64Kbps speech silence compression in atm network, the invention will be further elaborated.

This example is in ATM voice/data connecting system, adopts a cover silence compression equipment that 1024 road cluster voice signals of input are carried out silence compression, has realized the transmission of voice signal in atm network, and input speech channel number can be less than 1024.Silence compression equipment applied environment as shown in Figure 7.Wherein silence compression equipment is realized silence compression, and voice band business after the ATM connecting system will compress and data service are adopted AAL2 and the adaptive back of AAL5 to form ATM cell respectively and sent into atm network.At receiving terminal, each road voice signal of the at first professional and voice band business according to VPI (VPI) mask data in the Virtual Channel connection (VPC), and recovery process silence compression device processes is reformulated the cluster voice signal.

To the PCM voice signal of 64Kbit/s, silence compression device processes time delay is 0.125 * L millisecond, and interpolation gain is about 2.5.

In the adaptive mode of AAL2, according to ITU-TI.363.2, the ATM cell payload of 48 bytes is divided into little cell of regular length or variable-length, is called CPS (common part sublayer) grouping.Each CPS grouping has its LLID and length mark (LI).AAL2 (ATM Adaptation Layer type 2) is divided into SSCS (service specific convergence sublayer) and CPS.The adaptation procedure of CPS as shown in Figure 8.Fig. 9 has provided the definition format of AAL2 CPS packets headers and CPS_PDU (CPS protocol Data Unit) domain, and wherein RES is for keeping the position, and HEC is the packets headers error detection, and P is a parity check bit, and SN is cell sequence number counting.

As seen from Figure 8, STF is a domain, CID is a connection identifier, the AAL2 adaptation layer protocol can realize in an ATM connection that a plurality of LLC's (logical links connection) is multiplexing, because a CPS grouping may be crossed over cell boundaries, in the domain of CPS_PDU, be provided with an offset domain pointer (OSF), be used to refer to the original position of first complete packet in the current cell.When the cluster voice signal used AAL2 adaptive, because the requirement of delay character, the CPS block length must be fixed, thereby LI can do other purposes in the CPS packets headers.In order to assemble the cluster speech channel of maximum M=1024 time slots, need 10 bit CID sign indicating numbers, but CID has only 8 bits, transmit 10 bit time slots number sign and 3 bit background noise average powers altogether with 8 bit CID and 5 bit LI.Because silence compression equipment carries out one-time detection to each road signal L byte, for reducing overall delay, an AAL2 CPS packet payload length of field also is the L byte, and the wait time delay is τ=0.125 * L millisecond.The adaptive overall delay of silence compression equipment and AAL assembles the time delay sum for waiting for time delay, silence compression device processes time delay and AAL2.

The efficient after adaptive improves situation through silence compression device processes and AAL2 to analyze the cluster voice signal below.If atm network has N Service Access point (n ₁, n ₂..., n _N), every Service Access point inserts 1 road cluster voice signal, every road cluster voice signal has M speech channel, the telephone traffic of each speech channel is a Ireland (E), and etc. general one of N Service Access point (comprising the local service access point), the then n of arriving _iArrive n _jTelephone traffic be A=a * M/N (E), n _iThere is the k speech channel to arrive n _jProbability be: $P_k\left(a\right)=\frac{A^k}{k!}/{\mathrm{\Σ}}_{i=0}^M\frac{A^i}{i!}----\left(1\right)$

If forbid that business shared ratio in all speech channels of compressing is γ, the speech channel that allows compression is after quiet processing, and about 60% speech channel is in mute state, and then this k road speech total information rate after silence compression device processes and AAL2 are adaptive is: $R_k\left(\mathrm{\γ}\right)=\left|\frac{[\mathrm{\γk}+0.4k(1-\mathrm{\γ})(L+3)}{47}\right|\×53\×8/125L----(\mathrm{Mbit}/s)\left(2\right)$

Node n _iArrive node n _jThe assembly average of speed summation be: $R(a,\mathrm{\γ})=(N-1){\mathrm{\Σ}}_{k=0}^M{R}_k\left(\mathrm{\γ}\right)P_k\left(a\right)----\left(3\right)$

If M=1024, N=4, (a, γ) curve is as shown in Figure 9 for the R when L=6, different γ.A=0.1 (E), γ=0.1 o'clock, R (0.1,0.1)=16.93 (Mbit/s); A=0.1 (E), γ=1 o'clock, R (0.1,1)=34.76Mbit/s this shows, owing to utilized silence compression equipment, can transmit the 34Mbit/s signal with the bandwidth of 17Mbit/s.γ=1 o'clock is equivalent to not have the silence compression environment division.Efficient after silence compression equipment and AAL2 are adaptive is:

$=\frac{(N-1){\mathrm{\Σ}}_{k=0}^M\mathrm{kL}\×P_k\left(a\right)}{R(a,\mathrm{\γ})}----\left(4\right)$

M=1024, L=6, at single channel traffic carrying capacity a=0.05,0.1 and 0.2 o'clock, gross efficiency η is with the change curve of forbidding compressing professional ratio γ as shown in figure 10.For example, during a=0.1 E, γ=1 is equivalent to not have silence compression equipment, and efficient only is η (0.1,1)=0.56.After adopting silence compression equipment, efficient is improved significantly: as a=0.1E, and γ=0 o'clock, η (0.1,1)=1.3, when efficient is no silence compression equipment 2.32 times, promptly the bandwidth availability ratio of transmission line is brought up to original 2.32 times.

Analyze the band utilization situation below particularly.To 1024 road cluster voice signals,, then need to be not less than the 65Mbit/s transmission band if adopt the transmission of circuit common exchanged form; And after the silence compression device processes, send into atm network as the professional transmission of VBR, and at N=4, L=6, if γ=0.1 o'clock during single channel traffic carrying capacity a=0.1, needs the average bandwidth of 16.93Mbit/s.The advantage of the high efficiency of silence compression equipment and ATM statistic multiplexing here is fully reflected.Figure 11 has provided the probability that the bit rate that generates behind the cell surpasses a certain bandwidth B, and with γ=0.1, a=0.1E is an example, if require because the bandwidth deficiency causes the probability of cell loss concealment less than 1 * 10 ^-5, then the fixed-bandwidth with 24Mbit/s is enough to transmit the cell that forms after the silence compression device processes.During no silence compression equipment, the fixed-bandwidth that requires during a=0.1E is respectively 52.This means,, also can save the large volumes of channels resource even adopt the speech of the fixed-bandwidth transmission of Permanent Virtual Channel (PVP) through the silence compression device processes.

Claims (4)

1, a kind of multichannel 64Kbps speech silence compression method that is applicable to packet network, it is characterized in that: this method comprises data transmission procedure and DRP data reception process, wherein:

Data transmission procedure may further comprise the steps:

(A) initialization is made as state of activation with the status register of all time slots, activate counter, quiet counter zero setting continuously continuously, and set handling duration T presets quiet constant K, M, N, activates constant P and signal sample constant S;

(B) read timeslot number since 0;

(C) time slot that above-mentioned steps (B) is read is judged, if last time slot disposes, changes step (D) and continues operation, continues operation otherwise change step (E);

(D) wait for end, change step (B) then and continue operation to epicycle handling duration T;

(E) extract the compression sign of these data that read timeslot number;

(F) according to above-mentioned steps (E) but the compression sign judgment data that obtains silence compression whether as not allowing compression, change step (G) and continue operation, continue operation otherwise change step (H);

(G) add timeslot number before data, the data of this time slot of transparent transmission are changeed step (B) then and are continued operation;

(H) the state decision according to current these time slot data abandons or these time slot data of transparent transmission, changes step (B) then and continues operation;

DRP data reception process may further comprise the steps:

(a) according to timeslot number the data that receive are packed in the corresponding time slot of the corresponding frame of receiving terminal;

(b) fill the idle time slot of receiving terminal with white noise.

2, the multichannel 64Kbps speech silence compression method that is applicable to packet network according to claim 1 is characterized in that, described state decision according to current these time slot data abandons or these time slot data of transparent transmission are carried out according to following steps:

(A1) judge whether L signal sample in this handling duration T has S greater than thresholding,, forward step (A2) to and continue following operation, continue following operation otherwise change step (A3) if having;

(A2) carry out follow-up judgement and quiet processing;

(A3) the value R of read states register is a state of activation as R, changes step (A4) and continues operation, as R is mute state, changes step (A7) and continues operation, as the R mute state that is as the criterion, change step (A12) and continue operation,, change step (A13) and continue operation as the R state of activation that is as the criterion;

(A4) quiet counter adds 1 continuously;

(A5) do you whether judge quiet number of times=K? if be not equal to K, change step (A20) and continue operation, continue operation otherwise change step (A6);

(A6) the configuration state register mute state that is as the criterion is changeed step (A20) then and is continued operation;

(A7) quiet counter adds 1 continuously;

(A8) activate counter puts 0 continuously;

(A9) whether judge quiet number of times 〉=M,, change step (A10) and continue operation, continue operation otherwise change step (A19) as greater than M;

(A10) the configuration state register is a mute state;

(A11) quiet counter puts 0 continuously, changes step (A20) then and continues operation;

(A12) abandon to return after the data and handle next time slot;

(A13) quiet counter adds 1 continuously;

(A14) judge whether quiet number of times equals=N, as equating that changeing step (A16) continues operation, continues operation otherwise change step (A15);

(A15) activate counter puts 0 continuously, changes step (A20) then and continues operation;

(A16) the configuration state register is a mute state;

(A17) quiet counter puts 0 continuously;

(A18) activate counter puts 0 continuously, changes step (A20) then and continues operation;

(A19) return next time slot of processing after transmission timeslot number, noise average power and the data;

(A20) transmit timeslot number and data, return and handle next time slot.

3, the multichannel 64Kbps speech silence compression method that is applicable to packet network according to claim 2 is characterized in that, describedly carries out follow-up judgement and quiet processing is carried out according to following steps:

(C1) the value R of read states register is a state of activation as R, changes step (C16) and continues operation, as R is mute state, changes step (C2) and continues operation, as the R mute state that is as the criterion, change step (C8) and continue operation,, change step (C10) and continue operation as the R state of activation that is as the criterion;

(C2) activate counter adds 1 continuously;

(C3) judge that the activation number of times whether more than or equal to P, is not more than P if activate number of times, change step (C4) and continue to operate, continue operation otherwise change step (C5);

(C4) quiet counter adds 1 continuously, changes step (C15) then and continues operation;

(C5) the configuration state register is for activating;

(C6) activate counter puts 0 continuously;

(C7) quiet counter puts 0 continuously, changes step (C16) then and continues operation;

(C8) the configuration state register is for activating;

(C9) activate counter adds 1 continuously, changes step (C16) then and continues operation;

(C10) activate counter adds 1 continuously;

(C11) judge whether equal P, if equal, the step of walking around (C12) continues operation if activating number of times, otherwise step (C14) continues operation;

(C12) the configuration state register is for activating;

(C13) activate counter puts 0 continuously;

(C14) quiet counter puts 0 continuously;

(C15) return next time slot of processing after transmission timeslot number, noise average power and the data;

(C16) return next time slot of processing after transmission timeslot number and the data.

4, the multichannel 64Kbps speech silence compression method that is applicable to packet network according to claim 1 is characterized in that, described handling duration T, and quiet constant K, M, N activate constant P and signal sample constant S and determine by emulation according to system condition.

Images