ACOUSTIC ECHO CANCELLING METHOD
The present invention relates to a method for eliminating an acoustic echo that returns to a first mobile station from a second mobile station in a digital mobile communication system. The invention also relates to a mobile communication system comprising a first mobile services switching centre and a mobile station having a data transmission connection to the first mobile services switching centre, an echo canceller arranged on the data transmission connection between the first mobile services switching centre and the first mobile station, and a second mobile services switching centre and a mobile station having a data transmission connection to the second mobile services switching centre, and in which mobile communication system the first mobile station and the second mobile station communicate with each other via the first mobile services switching centre and the second mobile services switching centre. Furthermore, the invention relates to an acoustic echo canceller comprising echo cancelling means whose activation prevents an acoustic echo received by the echo canceller from propagating therefrom.
The present invention particularly relates to acoustic echo cancellation in a call between two mobile stations, although the invention is also suitable for use in other connections. Many mobile stations in use generate an acoustic echo audible to the other party of the call. The acoustic echo is generated when sound signals propagate during a call from the headphone of the mobile station to the microphone of the mobile station. Normally, this echo should be eliminated in the mobile station, but practice has shown that not all the mobile stations in use are capable of doing it. In a prior art solution, an acoustic echo canceller AEC is arranged in a mobile communication system network. Said echo canceller eliminates an acoustic echo generated by the mobile station from downlink speech and transmitted in the uplink transmission direction, i.e. from mobile station to mobile services switching centre. The prior art echo canceller monitors downlink speech activity. When detecting speech activity, the echo canceller starts after a predetermined delay replacing the uplink speech signals by comfort noise CN. Consequently, the acoustic echo transmitted from the mobile station cannot propagate past the echo canceller to other parts of the network.
It is a disadvantage of the prior art solution described above that it is not capable of preventing acoustic echo from being transmitted to the mobile station from other parts of the system. A situation of this kind may rise if a
first mobile station communicates with a second echoing mobile station located in the area of a mobile telephone network not having an acoustic echo canceller of the above-described type for use. In that case, the user of the first mobile station hears the acoustic echo that returns from the second mobile station. Naturally, a typical solution would be to also arrange an acoustic echo canceller in the mobile telephone network part in which the echoing mobile station is located. However, this is not always possible, because the second mobile station is possibly located in another operator's network, for instance.
An object of the present invention is to solve the above-described problem and provide a solution by means of which an acoustic echo can be eliminated more efficiently in a mobile communication system. This objective is achieved by the method of the invention, which is characterized in that the method comprises the steps of: arranging in the mobile communication system an echo canceller between a mobile services switching centre and the first mobile station, the echo canceller comprising means for eliminating an echo originating from the second mobile station, calculating a delay in an echo path, i.e. the time it takes for a speech signal to propagate from the echo canceller to the second mobile station and back to the echo canceller, and waiting for a period of time corresponding to the delay in the echo path from the moment the speech signals that originate from the first mobile station leave the echo canceller for the second mobile station, whereupon the echo canceller is activated to eliminate the speech signals having echoed from the second mobile station.
The invention also relates to a mobile communication system in which the method of the invention can be utilized. The mobile communication system of the invention, in which system an echo canceller is arranged on the data transmission connection between the first mobile services switching centre and the first mobile station, is characterized in that in that said echo canceller comprises: detector means for detecting a telecommunication connec- tion between the two mobile stations, echo cancelling means whose activation prevents a speech signal that echoes from the second mobile station from propagating to the first mobile station, and control means for activating the echo cancelling means in response to the recognition of an acoustic echo transmitted from the second mobile station. Furthermore, the invention also relates to an acoustic echo canceller which is characterized in that it further comprises: detector means arranged
to detect signalling between speech processing units of a telecommunication connection, and control means responsive to the detector means for activating the echo cancelling means in response to the recognition of an acoustic echo after the detector means have detected the signalling between the speech processing units.
The invention is based on the idea that an acoustic echo returning to a first mobile station from a second mobile station can be eliminated with an echo canceller placed in the network part in which the first mobile station is located. The echo canceller in question is able to check that a call between the two mobile stations is involved by monitoring signalling between speech processing units of the system, for instance. If it then detects that a telecommunication connection uses tandem free operation, it has to prevent the acoustic echo from propagating during the connection.
The presence of an acoustic echo can be checked by manners known per se, by calculating the cross correlation between the transmission directions (UPLINK/DOWNLINK), for example. If the maximum correlation is in that case found, then an acoustic echo is involved, and the maximum correlation point then indicates the delay in the echo path. A control unit of the echo canceller unit can activate the echo cancelling means after the cross correla- tion point of the signals between the transmission directions is detected when the identifying means recognize speech activity of the first mobile station. However, before activating the echo cancelling means, the control unit preferably waits for a period of time corresponding to the delay in the echo path from the recognition of the speech activity. In this context, the delay of the echo path refers to the time it takes for a sound signal to propagate from the echo canceller to the second mobile station and back to the echo canceller.
The most significant advantages of the solution of the invention are therefore that an acoustic echo can be eliminated more efficiently in a mobile communication network and that a mobile communication network operator, for example, can prevent an acoustic echo that originates from another operator's network from propagating to a mobile station located in the coverage area of the operator's network.
The preferred embodiments of the method and mobile communication system of the invention are disclosed in the attached dependent claims 2 to 4 and 6 to 11.
In the following, the invention will be described in more detail by way of example with reference to the accompanying figures, in which
Figure 1 shows a block diagram of a prior art mobile communication system, Figure 2 shows a block diagram of a first preferred embodiment of the mobile communication system of the invention,
Figure 3 shows a block diagram of the echo canceller AEC2 of Figure 2 , and
Figure 4 shows a flow diagram of a first preferred embodiment of the method of the invention.
Figure 1 shows a block diagram of a prior art mobile communication system. The mobile communication system shown in Figure 1 can be a GSM system (Global System for Mobile Communications), for instance.
In the case of Figure 1 , a call is in progress between mobile stations MS1 and MS2. When the user of the mobile station MS2 speaks to a microphone 1 , the mobile station MS2 encodes the speech by a RPE-LTP speech encoding method used in the GSM system. The mobile station MS2 transmits the encoded speech via a radio path to a base station BTS2 from which it is further transmitted via a base station controller BSC2 to a speech processing unit TRAU2 (Transcoder/ Rate Adapter Unit).
The speech processing unit TRAU2 decodes the speech signals received from the mobile station MS2 and forwards them in a packed form, i.e. companded as PCM samples which are encoded according to the A / μ law, for example (described in more detail for instance in the ITU-T recommenda- tions G. 711 ) over a PCM connection via mobile services switching centres MSC2 and MSC1 and an acoustic echo canceller AEC to a speech processing unit T.RAU1. The speech processing unit TRAU1 encodes the speech originating from the mobile station MS2 using the RPE-LTP speech encoding method. Since a telecommunication connection between the two mobile stations is involved in the case of Figure 1 , tandem free operation TFO is in use therein. In other words, tandem free operation is used in the case of Figure 1 in order to avoid degraded speech quality because of encoding the RPE-LTP- encoded speech that originates from the mobile station MS2 into a format used by the mobile services switching centres (the format corresponding to speech coding used in a fixed PSTN network) and thereafter re-encoding it by
means of the same RPE-LTP speech encoding. This means that speech, which is received from the mobile station MS2 and encoded by the RPE-LTP speech encoding method, is transmitted from the speech processing unit TRAU2 to the speech processing unit TRAU1 on a subchannel that is formed of one or two least significant bits of the PCM samples. No transcoding operations (decoding and encoding) are executed to the speech in the two speech processing units. The receiving speech processing unit primarily selects for use the RPE-LTP-encoded speech which is transmitted on the subchannels in question. In the case of Figure 1 , tandem free operation is generated in connection with call establishment. In that case, the speech processing units TRAU1 and TRAU2 have generated tandem free operation in a manner known per se by mutual signalling. Publication WO96/32823, for example, describes tandem free operation in more detail. The speech processing unit TRAU1 forwards the RPE-LTP- encoded speech originating from the mobile station MS via a base station controller BSC1 and a base station BTS1 to the mobile station MS1 , the user of the mobile station MS1 then hearing the speech of the mobile station MS2 user through a speaker 2. In the case of Figure 1 , it is assumed that the mobile stations MS1 and MS2 participating in the call both generate an acoustic echo, i.e. sound signals are passed from their speakers 2 to their microphones 1. An acoustic echo canceller AEC1 is arranged between the mobile station MS1 and the mobile services switching centre MSC1. When detecting downlink speech ac- tivity (from the mobile services switching centre MSC1 to the mobile station MS1 ), the acoustic echo canceller becomes active in order to prevent the uplink speech signals from passing. Consequently, although an acoustic echo is generated in the mobile station MS1 , speech initially originating from the mobile station MS2 cannot return to the mobile station MS2, since the echo canceller AEC1 prevents the speech signals from propagating in the uplink direction. To prevent the mobile station MS2 user from noticing that the echo canceller AEC1 is active, the echo canceller AEC1 generates comfort noise (or alternatively, the echo canceller AEC1 directs the speech processing unit TRAU2 to generate comfort noise). Comfort noise means here noise with the same spectral properties and intensity as noise that prevails in the surroundings of the mobile station MS1 at a given time. Similar comfort noise is utilized
in the GSM system in connection with discontinuous transmission DTX of a mobile station, for example. Finnish Patent Application 971680, for example, discloses how an acoustic echo can be replaced by comfort noise that is transmitted to the mobile station when tandem free operation is used. Consequently, in the case of Figure 1 , the echo canceller AEC1 prevents the acoustic echo that returns from the mobile station MS1 from propagating by replacing the echo by comfort noise. On the other hand, no echo canceller is arranged on the data transmission connection between the mobile services switching centre MSC2 and the mobile station MS2, for which reason speech initially originating from the mobile station MS1 echoes back from the mobile station MS2 up to the user of the mobile station MS1. Since it is assumed in the case of Figure 1 that mobile telephone networks 1 and 2 belong to different operators, the operator of the network 1 cannot interfere in the situation in order to solve the problem. The decision whether an echo can- celler is installed in the mobile telephone network 2 is totally up to the operator of the network in question.
Figure 2 shows a block diagram of a first preferred embodiment of the mobile communication system of the invention. The block diagram of Figure 2 corresponds to the solution presented in Figure 1 in other respects ex- cept that an echo canceller AEC2 is added in accordance with the invention between the mobile services switching centre MSC1 and the mobile station MS1 in the system of Figure 2. The echo canceller AEC2, the structure of which is described in more detail in connection with Figure 3, prevents sound signals, such as speech signals, that initially originate from the mobile station MS1 from returning as an acoustic echo to the mobile station MS1 from the mobile station MS2.
As distinct from the case of Figure 2, the echo canceller AEC2 may also be located elsewhere. Consequently, it may be located in connection with the mobile services switching centre MSC1 , for instance, or it may even be integrated in the mobile station MS1.
Figure 3 shows a block diagram of the echo canceller AEC2 of Figure 2. In the case of Figure 3 it is assumed that the echo canceller is installed in the location presented in Figure 2, whereby it communicates with the mobile services switching centre MSC1 and the echo canceller AEC1. The echo canceller AEC2 comprises a detector 3 monitoring the signalling between the speech processing units TRAU1 and TRAU2 in con-
nection with call establishment between the mobile stations MS1 and MS2. The signalling in question, used by the speech processing units TRAU1 and TRAU2 for agreeing on tandem free operation TFO, may be effected by utilizing the least significant bits of the PCM connection between the speech proc- essing units TRAU1 and TRAU2 and the mobile services switching centres MSC1 and MSC2, for example. Consequently, the detector 3 monitors the least significant bits in question to find out if the call is a call between two mobile stations. This is indicated, for example, by the speech processing units TRAU1 and TRAU2 agreeing on tandem free operation by signalling in which the least significant bits in question are utilized.
After the detector 3 has detected the call between the two mobile stations, a control unit 6 of the echo canceller AEC2 activates a counter 4 to calculate a cross correlation for the uplink and downlink speech signals. In that case, the maximum correlation point indicates the delay in the echo path. The delay in the echo path is the period of time from the moment a sound signal, such as speech, originating from the mobile station MS1 passes the echo canceller AEC2 to the moment the acoustic echo generated from the speech in the mobile station MS2 returns to the echo canceller AEC2. If the counter 4 does not in that case find the maximum correlation, it indicates that no acous- tic echo exists, and, consequently, there is no need to activate the echo canceller AEC2.
When the delay in the echo path is known, the echo canceller AEC2 monitors uplink speech signals via a speech activity identifier 5. When detecting speech activity in the mobile station MS1 , i.e. speech originating from the mobile station MS1 , the identifier 5 informs the control unit 6 thereof. The control unit 6 waits for a period of time corresponding to the delay in the echo path from the moment in question, whereupon it activates echo cancelling means 8 to prevent the downlink speech signals from propagating to the mobile station MS1. At the same time, it activates a signal generator 7 to generate comfort noise that is transmitted through the echo cancelling means 8 to the mobile station MS1.
As distinct from Figure 3, in connection with the activation of the echo cancelling means, the control unit 6 of the echo canceller AEC2 can send a pre-determined control message to the speech processing unit TRAU1 located between the control unit and the mobile station MS1. The control mes-
sage makes the speech processing unit generate and transmit the above- mentioned comfort noise to the mobile station MS1.
After activating the echo cancelling means 8, the control unit 6 keeps them preferably activated until the identifier 5 detects that the speech activity of the mobile station MS1 has terminated, whereupon it waits for a period of time corresponding to the delay in the echo path before deactivating the echo cancelling means 8. In addition, the identifier 5 is preferably arranged to also monitor downlink speech activity. Consequently, it is capable of detecting a double-talk situation, i.e. a situation in which both the user of the mobile sta- tion MS1 and the user of the mobile station MS2 talk simultaneously. It can detect the double-talk situation when the signal level of the downlink sound signals exceeds a given threashold during the speech activity of the mobile station MS1. The identifier 5 then informs the control unit 6 of the double-talk situation, whereupon the control unit 6 deactivates the echo canceller. Figure 4 shows a flow diagram of a first preferred embodiment of the method of the invention. The block diagram of Figure 4 can be utilized in a call between the mobile stations MS1 and MS2 of Figure 3, for example, from call establishment onward.
In block A, signalling between speech processing units is moni- tored. If it is then detected in connection with call establishment, for example, that the speech processing units agree upon tandem free operation by signalling, then a call between two mobile stations is involved. In that case, the routine proceeds via block B to block C.
In block C, a delay in an echo path is calculated, i.e. the period of time it takes until the speech signals return in the form of an acoustic echo. The delay in the echo path can be found out, for example, by calculating a cross correlation between the transmission directions, the maximum correlation point then indicating the delay. If no distinct maximum correlation is found, this means that no acoustic echo exists, in which case there is also no need to activate echo cancellation.
In block D it is checked if the call has terminated. If not, the routine proceeds to block E. In block E, the speech activity of the mobile station MS1 is monitored. When speech activity is detected, the routine proceeds to block F. If no speech activity is detected, the routine proceeds at regular intervals to block D to check if the call has terminated.
In block F, it is necessary to wait for a period of time corresponding to the delay in the echo path, whereupon an echo canceller is activated in block G.
In block H it is checked if a double-talk situation exists, i.e. if the us- ers of the mobile stations participating in the call both talk simultaneously. If not, it is next checked if the speech activity of the mobile station MS has terminated in block I. In other words, if the period of time from the last recognized speech activity of the mobile station MS1 exceeds the time corresponding to the delay in the echo path, it indicates that all echo signals generated from the speech signals of the mobile station MS1 have been eliminated.
The blocks H and I are re-checked until one or the other condition is fulfilled, whereby the routine proceeds to block J where the echo canceller is deactivated.
The routine returns to block D from block J, where it is checked if the call is terminated. If not, the blocks of the flow diagram are gone through again from block E onward until it is finally found in block D that the call is terminated.
It is to be understood that the above description and the related figures are only intended to illustrate the present invention. Consequently, the present invention can also be utilized in systems other than the GSM system described above by way of example. Different variations and modifications will be apparent to those skilled in the art without deviating from the scope and spirit of the invention disclosed in the attached claims.