KR100770819B1 - Method and system for performing discontinuous transmission in a communications system - Google Patents

Abstract

The technique for fast in-band signaling and discontinuous transmission (DTX) of configuration changes and protocol messages in a voice communication system provides cost efficiency in terms of wireless transmission capacity, in terms of fixed line transmission and in terms of implementation effort. to provide. An example method for performing discontinuous transmission (DTX) in a communication system, in which source data is interleaved for transmission from a first component in the communication system to a second component in the communication system, detects periods of source data inactivity state. And transmitting a Silence Descriptor (SID) frame from the first component to the second component during the source data inactivity period, wherein some of the transmitted SID frames use different algorithms than those used for the source data. Interleaved. For example, the source data may be block diagonally interleaved and some of the SID frames may be block interleaved. An example method of performing configuration change in a communication system includes transmitting an extended frame instead of a voice data frame, the extended frame comprising a total bit pattern for distinguishing between the voice data frame and the extended frame and displaying a configuration change indication. To pass. The extended frame further includes a data field for indicating a specific configuration change to be made. For example, if the communication system is an AMR system, the extended frame may be used to change the active codec mode set. Optionally, the extended frame can be used to change the phase of the codec information.

Communication systems, voice data frames, silence descriptor (SID) frames, extended frames, source data, discontinuous transmission

Description

METHOD AND SYSTEM FOR PERFORMING DISCONTINUOUS TRANSMISSION IN A COMMUNICATIONS SYSTEM}

This application claims the privilege of US Provisional Application No. 60 / 109,694, filed November 24, 1998, the entire contents of which are incorporated herein by reference.
TECHNICAL FIELD The present invention relates to communication systems, and more particularly, to discontinuous transmission (TDX) and configuration changes in an adaptive multirate communication system.

Today, multi-mode coding systems can be used using at least two different source and channel codec modes to maintain near optimal communication quality under variable transport channel conditions. For bad channels, a mode with a low source coding bit rate and a high channel error protection may be selected. On the other hand, in the case of a good channel, a codec mode having a high source coding bit rate and a relatively low channel error protection can be selected.

As is known in the art, such a multi-mode coding system must convey (explicitly or implicitly) the codec mode actually selected to the receiving decoder to enable proper decoding of the received data. Bidirectional communication systems with codec mode adaptation must additionally transmit similar information over the return link. This is either quantized link measurement data describing the current forward channel state or a corresponding codec mode request / command with the channel state account. Such link adaptation data is known in the art as codec mode information and includes a codec mode indication (actually selected codec mode) and a codec mode request / command (codec mode used at the transmitting side). The Evolving Global System for Mobile Communication (GSM) Adaptive Multi-Rate (AMR) standard has the codec mode adaptation described above.

In such AMR systems, in-band signaling is used to reallocate portions of voice transmission resources for transmitting control information. It applies where no other suitable control channel is available. The GSM AMR speech coding standard is an example of using in-band signaling. It uses part of the GSM voice traffic channel for the transmission of AMR link adaptation data. More specifically, the GSM AMR standard provides an in-band channel for the transmission of codec mode information.

Codec mode information consists of a codec mode request / command and a codec mode indication, which are alternately transmitted every second frame (every 40 ms). The codec mode information identifies the codec mode within a subset of up to four codec modes from eight (adaptive full-rate voice, AFS) or six (adaptive half-rate voice, AHS) available modes. This codec mode subset is called the active codec set.

In any communication system, including the GSM AMR system described above, the transmission capacity is a limited and expensive resource. For this reason, in order to save transmission capacity, discontinuous transmission (DTX) is widely applied in voice transmission. Sometimes DTX is called Voice Driven Transmission (VOX). The basic principle of DTX is to turn off transmission while voice is inactive. Instead, so-called comfort noise (CN) parameters, which are usually a kind of background noise, are transmitted so that the decoder can reproduce inactive signals. CN parameters require smaller transmission resources than voice. DTX is also an important feature for mobile phones because it allows the power consuming device (eg, a wireless transmitter) to be turned off while it is inactive. Doing so can save battery power and increase phone call time.

In a two-way communication system using DTX, one link will generally be active (because one caller is listening if the other is speaking) while the other will be inactive. The active link will result in some reduction in the frame rate, which must convey a Silence Descriptor (SID) frame (known as background information, or comfort noise, descriptor frame) to the receiver. The SID frame contains the CN parameter and, for example, allows the receiver to generate a comfortable noise silence signal so as to reassure the listening party that the connection is still active.

In the current GSM voice coding standards FR, HR and EFR, DTX is implemented in a very similar manner. By way of example, the state of the art in voice communication operating with DTX in a GSM system will be described with respect to the GSM EFR codec. For further information, see, for example, the GSM 06.11, GSM 06.12, GSM 06.21, GSM 06.22, GSM 06.31, GSM 06.41, GSM 06.61, GSM 06.62, GSM 06.81 standards, and related documents. The GSM EFR scheme has the following characteristics:

The end of the voice activity state is indicated by the transmission of the first SID frame, which is not phase aligned with the SACCH. Rather, it comes immediately after the final activity speech frame. Following such a first SID frame, update SID frames are transmitted in one cycle (= 480 ms) per 24 frames. Update SID frame transmissions occur within the wireless subsystem and are aligned with a time alignment flag (TAF) derived from the SACCH frame structure. Unlike SID frames, no other frames are transmitted during inactivity. Simply resuming the transmission of the active speech frame ends the period of inactivity.

RSS treats SID frames as regular speech frames. This means in particular that the same channel coding and diagonal interleaving are used for speech frames. In the pleasant noise parameter describing the spectral shape and gain of the inactive signal, a number of effective 43 net bits are used. 95 net bits are used in a special SID bit pattern for identifying a frame as an SID frame and distinguishing it from a voice frame. CN parameters are encoded differently than parameters derived from the last transmitted speech frames.

The SID frame transmission described above is described for TCH / FS (ie traffic channel / full-rate voice) in FIG. 1 and for TCH / HS (ie traffic channel / half-rate voice) in FIG. 2. The upper row represents the speech frame as shown at the input of the speech encoder. The middle row shows a TDMA frame that transmits each voice or SID bit via the air interface. The lower row shows the speech or pleasant noise frame following the speech decoder. All speech frames are exactly 20ms long. TDMA frames are on average exactly 5ms apart. TDMA frames for SACCH and IDLE are not shown. Implementation delays and other side effects are also not shown.

Unlike regular transmission of SID frames that are time aligned in synchronization with a fixed time structure, ITU-T Recommendation G.729 / Annex B proposes a DTX method that transmits SID frames whenever an update of CN parameters is required. This is because the SID frame has changed considerably since the last SID frame transmission.

For known Pacific Digital Cellular (PDC) systems with VOX capability, special trailing and introductory frames are used to indicate transitions from speech to inactivity or from inactivity to speech, respectively (e.g., RCR STD -27D). These frames contain unique bit patterns for identifying them on the gross bit level. The trailing frames consist of a first channel frame that does not contain any information other than an identification bit pattern and a second channel frame that contains comfortable noise parameters describing the inactive signal. During voice inactivity, trailing frames are transmitted periodically to enable the receiving side to update the generation of pleasant noise. The same interleaving is used for speech frames in both the trailing and opening frames.

The conventional DTX solutions described above implemented with GSM FR, EFR, and HR are not very suitable for use in multi-mode coding systems. This is a result from the fact that SID frame signaling is performed at the net bit level. The special bit pattern that identifies the SID frame is part of the net bit stream. The SID frame detection unit at the receiver is executed after de-interleaving and channel decoding. This approach is inadequate for multimode coding systems having more than one source and channel mode because SID frame identification depends on the correct selection of codec mode for channel decoding. Due to the possibility of mode transmission error, the correct codec mode at the receiver may not always be guaranteed.

In addition, for similar reasons, changing the interleaving scheme for either different codec modes or SID frames is also impractical for reasons of complexity. This approach requires, in addition to speech frame deinterleaving and channel decoding, in the worst case, deinterleaving of the SID frame, even worse, channel decoding.

In addition, there are at least two important issues in adopting a PDC implementation. First, since the trailing frame consists of two traffic frames, inactive transmission mode is relatively inefficient in terms of transmission power savings. Each pleasant noise parameter update requires transmission of two frames. Second, the transition from the voice inactive state to the active state is indicated by introductory frames and either of the portions of the speech initiation can be truncated or the transmission of the speech initiation can be resumed with a delay by the introductory frame. The former effect directly degrades the quality of the reconstructed speech, while the latter increases the speech transmission delay, which can cause degradation of the quality of the conversation.

Note that applying a common diagonal interleaving scheme currently being implemented in both GSM and PDC for two frames and voice frames for the SID creates an additional problem. Applying diagonal interleaving for the transmission of a single SID frame means that only half of all transmitted TDMA frames contain SID information, while the other half remains unused and wasted (this wasted half). Bursts are shown in Figures 1 and 2), which are inefficient in terms of radio resource usage and power consumption.

This loss of efficiency in current GSM and PDC systems is small because of relatively little SID frame transmission. However, it is very severe for new multi-mode communication systems with codec mode adaptation. In order to increase the adaptive performance, it is necessary to transmit information (adaptation data) more frequently through the inactive link than the transmission of the SID frame in the current system.

In addition, there is a certain upper limit of radio channel activity during inactivity (eg, AMR system requirements are: TCH / AFS: 16 TDMA frames per 480 ms multiple frames; TCH / AHS: 12 TDMA frames per 480 ms multiple frames). to be). Wasting half of the available radio resources means that codec mode information can in principle be transmitted at half the frequency of what is possible. As a result, late codec mode adaptation results in potential performance loss.

An additional problem in applying the same diagonal interleaving to SID frames (including codec mode information) for speech frames is that delays are caused by this kind of interleaving. In achieving the maximum possible performance of codec mode adaptation in a multi-mode communication system, the transmission delay of codec mode information should be kept to a minimum. This disables diagonal interleaving.

A particular problem in systems with DTX is the detection of the voice initiation after an inactive state cycle. If you miss the start, the audio output of the decoder will be cut off. On the other hand, if an untransmitted frame is mistakenly detected as a voice initiation frame, unwanted pops or bangs can be generated that can significantly degrade the quality of the communication.

In principle, AMR systems driven by DTX simply need to send a codec mode request for the current active link over the inactive link. Codec mode indications for inactive links need not be sent. However, when the inactive link becomes active again, the appropriate codec mode must be selected. A solution to select a codec mode for speech initiation after inactivity should be found to ensure that the sending and receiving sides use the same mode. This codec mode must also be suitable for the current radio channel conditions.

Apart from codec mode signaling in the AMR standard, no additional fast control channel is available so far. However, fast configuration changes (e.g., change the set of active codecs, change the phase of the codec mode information for the purpose of minimizing transmission delays, hand over to existing GSM codecs such as FR, EFR, or HR, and / or Or such a channel to enable the transition to future applications such as wideband codecs, voice and data, or multiple media.

Thus, there is a need for improved methods and apparatus for implementing DTX and configuration changes in adaptive multirate systems.

The present invention satisfies the above and other requirements by providing a novel solution for DTX, fast in-band signaling of configuration change and protocol messages, and interaction of the two operations in an adaptive multirate system. Effectively, the methods and apparatus described above are cost effective in terms of wireless transmission capacity, in terms of fixed line transmission, and in terms of implementation effort.

An example method for performing discontinuous transmission (DTX) in a communication system in which source data is interleaved for transmission from a first component in a communication system to a second component in the communication system may include periods of source data inactivity. Detecting, and transmitting a Silence Descriptor (SID) frame from the first component to the second component during the source data inactivity period, wherein the particular SID frame transmitted is a different algorithm than that used for the source data. Are interleaved using.

The example method sends a SID frame of a first type to indicate a transition from a source data inactive state to a source data inactive state, and periodically transmits a SID frame of a second type during the source data inactive state. And transmitting a SID frame of the third form to indicate a transition from the source data inactive state to the source data active state. Effectively, if the communication system is an adaptive multi-rate (AMR) system, the SID frame may include codec mode information in addition to the silence description information.

An example method of transmitting a protocol message from a first component to a second component in a voice communication system includes sending an escape frame in place of a voice data frame, the extended frame being in conjunction with the voice data frame. Contains a total bit pattern to distinguish extended frames and carries a protocol message. This extended frame may further include a data field for indicating a specific protocol message to the second component.

An example method of achieving a configuration change in a communication system includes transmitting an extended frame instead of a voice data frame, which includes a total bit pattern and distinguishes the configuration change indication to distinguish the extended frame from the voice data frame. To pass. The extended frame may further include a data field to indicate to the second component a specific configuration change to be made.

For example, if the communication system is an AMR system, the extended frame may be used to change the set of active codec modes. Optionally, an extended frame can be used to change the phase of the codec information.

The above-described configuration and effects of the present invention and other configurations and effects will be described in detail below with reference to the embodiments shown in the drawings. Those skilled in the art will appreciate that the described embodiments are provided for purposes of illustration and understanding and that many equivalent embodiments can be anticipated herein.

1 is a diagram illustrating an example full-rate silence descriptor (SID) frame transmission scheme according to the prior art.

2 is a diagram illustrating an exemplary half-rate silence descriptor (SID) frame transmission scheme according to the prior art.

3 illustrates an example adaptive multirate communication system that may implement the present invention.

4 is a diagram illustrating an exemplary SID frame format according to the present invention.

5 illustrates an example full-rate SID frame interleaving scheme in accordance with the present invention.

6 illustrates an exemplary half-rate SID frame interleaving scheme in accordance with the present invention.

7 is a diagram illustrating an exemplary first SID frame format according to the present invention.

8 illustrates an exemplary speech initiation frame format in accordance with the present invention.

9 is a diagram illustrating an exemplary scheme for prohibiting a first SID frame according to the present invention.

10 is a diagram illustrating an example scheme for prohibiting regular SID frames in accordance with the present invention.

FIG. 11 illustrates an example full-rate scheme for detecting a transition from a voice inactive state to a voice active state, in accordance with the present invention.

12 is a diagram illustrating an exemplary half-rate configuration for detecting a transition from a voice inactive state to a voice active state, in accordance with the present invention.

FIG. 13 illustrates an example full-rate scheme for detecting a voice initiation when a voice-initiation indication frame is replaced with a system configuration change frame according to the present invention.

14 is a diagram illustrating an exemplary half-rate scheme for detecting a voice initiation when a voice-initiation indication frame is replaced with a system configuration change frame according to the present invention.

Although embodiments of the present invention will be described for voice transmission in a GSM system, those skilled in the art will readily recognize that the disclosed technology may equally apply to other content. For example, the present invention is readily applied in any wireless or fixed line communication system, such as TDMA systems (eg, D-AMPS), PDC, IS95, and the Internet.

3 illustrates an example AMR communication system that may implement the techniques of the present invention. The illustrated AMR system includes a transcoding and rate adaptation unit (TRAU) and a base station (BTS) on the network side, as well as a mobile station (MS). On the network side, a voice encoder (SPE) and a channel encoder (CHE), as well as a channel decoder (CHD) and a voice decoder (SPD), are connected via a well known serial A-bis interface. In each link, quality information is obtained by evaluating the current channel condition. Based on the channel state and in view of the possibility of restriction from the network controller, the codec mode controller arranged on the network side selects the codec mode to be applied.

The channel mode used (TCH / AFS or TCH / AHS) is controlled by the network. Uplink and downlink always use the same channel mode. In the case of codec mode adaptation, the receiver measures the link quality of the incoming link. This measurement is processed to yield a quality indicator. In the case of uplink adaptation, the quality indicator is supplied directly to the UL mode control unit. This unit compares the quality indicator with a certain threshold and generates codec mode commands that indicate the codec mode to be used on the uplink, taking into account possible constraints from the network control. This codec mode command is transmitted in-band to the mobile side where incoming voice signals are encoded in the corresponding codec mode.

In the case of downlink adaptation, the DL mode request generator in the mobile compares the quality indicator with a certain threshold to generate a codec mode request indicating the desired codec mode for the downlink. The codec mode request is sent in-band to the network side, where it is supplied to the DL mode control unit. This unit generally accepts the requested mode. However, taking into account possible constraints from the network control unit, the request can be invalidated. The final codec mode is used to encode incoming voice signals in the downlink direction.

In both uplink and downlink, the currently used codec mode is transmitted in-band as codec mode indication along with the coded speech data. At the decoder, the codec mode indication is decoded and used for decoding the received voice data.

Codec mode selection is performed among a set of codec modes, which may include 1 to 4 AMR codec modes. One to three switching thresholds and hysteresis lists used by the DL mode request generator and the UL mode control unit are associated with this set to generate codec mode requests and codec mode commands. These configuration parameters (ACS, threshold, hysteresis) are defined at call setup and can be modified at handover or during a call.

According to the present invention, the DTX in a system such as that shown in FIG. 3 is based on in-band signaling in the form of three different frames: SID_FIRST, regular SID, and voice start frame. These three frame types are the same in that they use a special total bit pattern to identify each frame. Furthermore they can carry payload data consisting of CN parameters and codec mode information. For example, implementations according to the invention, GSM 05.03: digital cellular telecommunications system (phase 2+); Channel coding (Draft ESTI EN 300 909 V7.2.0 (1999-11)) and GSM 06.93: digital cellular telecommunication system (phase 2+); See Discontinuous Transmission (DTX) for Adaptive Multi Rate (AMR) Voice Traffic Channel (Draft ESTI EN 301 707 V.7.2.0 (1999-11)). The entire contents of each of these are referred to herein.

SID frames are identified at the total bit level. SID frames are defined as being transmitted using k TDMA frames, that is, they consist of k * 114 bits. A suitable choice of k is four. In this case, the SID frame consists of 456 bits, that is, one channel frame of 456 bits for TCH / AFS and two channel frames of 228 bits for TCH / AHS. Each SID frame has an SID frame identification field and two message fields that contain a unique bit pattern. One message field is for channel encoded comfort noise (CN) parameters and the other field is for channel encoded codec mode information. The codec mode information field may include only a codec mode request or may be subdivided into two parts, two parts, one part containing a codec mode request / command, and the other part including a codec mode indication.

One embodiment of a regular SID frame format definition is given in FIG. 4. In this embodiment, the SID frame is composed of a 212-bit SID frame identifier, a 212-bit field for comfortable noise parameters, and a 32-bit field for codec mode information. In this embodiment, it is assumed that the CN parameter is encoded in a convolution manner and the codec mode information is composed of request / commands and instructions encoded in block encoding. In another solution, for example, if both the CN parameter and codec mode information are encoded using the same convolutional or block code, the two message fields may be put together.

According to the present invention, regular SID frames are block interleaved rather than diagonally. Although this may lose the possible interleaving gains (ie, the transmission is potentially less robust against transmission errors), SID frames generally contain less information than regular speech frames and therefore are more than those used for voice transmission. It can be protected using strong channel codes. This will either compensate for the interleaver gain loss or make the SID frame transmission more robust than is possible for current solutions (FR, EFR, or HR). Important information such as codec mode information and the like can be protected by, for example, stronger channel codes (relative to in-band transmission of codec mode information in regular speech frames). In addition, CN parameters are usually represented by much less bits than voice parameters. Thus, fewer CN bits can be protected with a lower rate channel code. As an example, all of the 35 CN bits can be protected, firstly by a 14-bit CRC code (which enables very robust error detection), and then by rate 1/4 convolutional code ( Can be protected by using the length constraint k = 5). Moreover, both CN parameters and codec mode information are generally relatively slowly changing information. Also, considering the proposed SID frame rate (every eighth frame), which is much higher than in existing solutions, the loss in case of SID frames due to channel error is also acceptable.

As shown in Figures 5 and 6, respectively, in both TCH / AFS and TCH / AHS, SID frames consisting of 4 * 114 bits are mapped according to the present invention by block interleaving into 4 TDMA frames. The purpose of the interleaver is to distribute SID frame bits to available TDMA frames in such a way that robustness to transmission errors is maximized. The diagonal interleaver for speech frames is not used. Since deinterleaving does not require complexity, this solution with a specific block interleaver for SID frames is feasible. In the worst case, the decoder performs both SID frame block deinterleaving and conventional speech frame diagonal deinterleaving, but there is at most one such decoder. Effectively, the problem in current GCM and PDC systems with wasted bits in TDMA frames belonging to SID frames is thus solved.

For TCH / AFC, the actual block interleaving scheme for SID frames is relatively insignificant. In order to obtain the maximum interleaver gain, the identification marker bits as well as the CN and codec mode information bits are distributed as equally as possible over the TDMA frame used for transmission.

In the case of TCH / AHS, special cases may occur due to the fact that the SID frame is transmitted using a two channel frame. This situation may occur when the first half of the TDMA frame containing the SID frame is transmitted and the second half cannot be transmitted due to voice initiation, as described in detail below with respect to the SID inhibited frame. In this case, it is important to be able to inhibit the SID pattern already transmitted. This is ensured by sending a second half of the pattern bit on the odd positions of the second half of the TDMA frame. Regarding codec mode information, it is important that a codec mode to be used for decoding the speech initiator is available. This can also be ensured by sending a second half of the codec mode indication bit on the odd position of the second half of the TDMA frame.

A possible solution is to map the pattern bits and codec mode indication bits on the TDMA frame by using diagonal interleaving. Thus, the CN bit and codec mode request / command bit are sent in odd positions of the first half of the TDMA frame and in even positions of the second half of the TDMA frame. The interleaving scheme described for the SID frame for TCH / AHS is described in FIG.

According to the present invention, a particular SID_FIRST frame is sent immediately after the last voice frame when going from active state to inactive state, the solution being to only identify the end of speech rather than sending CN parameter. One example solution for TCH / AFS is to use a 228 bit field consisting of 212 marker bits and 16 bits for codec mode information, as shown in FIG. Codec mode information, in turn (if a voice frame has been sent), is either a request / command or an indication. Therefore, the type of codec mode information transmitted by the SID_FIRST frame depends on the transmission step and the number of frames of the codec mode information. A particular interleaver maps the SID_FIRST frame onto the available 228 bits in an unused half burst. 5 shows the above-described transmission scheme of the SID_FIRST frame for TCH / AFS. Note that there is no longer wasted half burst.

A similar solution for TCH / AHS is to send the SID_FIRST identification pattern and codec mode information on two available and unused half bursts. However, one example of making the detection of SID FIRST more reliable is also to use the next two TDMA frames. This means that two channel frames SID_FIRST_1 and SID_FIRST_2 are transmitted. If possible, the same 228-bit frame (consisting of 212 marker bits and 16 bits for codec mode information, see FIG. 7) contains the final speech frame (unused half burst) if the 228-bit frame is currently used in the TCH / AFS example solution. On even positions of the TDMA frame and on odd positions of two consecutive TDMA frames. This kind of diagonal mapping allows the use of existing diagonal (de) interleavers. The codec mode information is either a request / command or an indication according to the transmission step and the number of frames of the codec mode information. If voice was transmitted, that kind of codec mode information that would have been transmitted in each channel frame is transmitted. This mapping is performed in such a way that the same part of both the pattern bit and the codec mode information is placed in the first two and the second two of the used TDMA frames.

6 illustrates a technique for further increasing the reliability of SID_FIRST frame detection. According to the invention, even positions of two further TDMA frames are filled with an additional identification pattern. It is also possible to use some of these half bursts in the transmission of codec mode information. If so often repeats that all available bits are used, the identification pattern may also be a code word of codec mode information. For example, if 114 bits are available and the code word for the codec mode information is 16 bits wide, it may be repeated 114/16 times.

Diagonal interleaving used in speech frames suggests that the odd positions of the first half of the TDMA frame containing the first speech frame after the inactivity state cycle are free for other purposes. According to the present invention, one solution to improving the start detection is to fill these bits with a specific start identification pattern. Moreover, portions of these bits may also be used to transmit a codec mode indication that signals the codec mode in which the first speech frame is encoded accordingly. As illustrated in FIG. 8, a solution to conveying both the start bit pattern and the codec mode indication is to repeat the codec mode indication code word to the extent that all available bits are used. One example for TCH / AFS is to repeat a 16-bit instruction code word 228/16 times. In the case of TCH / AHS, the 16-bit code word is repeated 114/16 times. This start frame is mapped by a particular interleaver on other unused half bursts. Respective frame transmission schemes for both TCH / AFS and TCH / AHS are shown in FIGS. 5 and 6.

For TCH / AHS, regular SID frames and SID_FIRST frames are transmitted using two channel frames. Thus, situations may arise where a higher priority voice initiator is transmitted after the first channel frame of the SID frame is transmitted, but before the second channel frame is transmitted. In such a case, even if it actually only receives its first half, an error incident may occur where the receiver does not detect the beginning and instead detects the SID or SID_FIRST frame.

To avoid this problem, when a first half of the TDMA frame containing SID_FIRST is transmitted but the second half cannot be transmitted due to voice initiation, a particular SID_FIRST inhibit frame is used instead of the regular initiation frame. The pattern bits belonging to the second half of the SID_FIRST frame that could have been transmitted are now inverted. This prevents the detection of the entire SID_FIRST pattern at the receiver. The codec mode information bits remain the same as from the original SID_FIRST frame. The receiver will get a frame that is unavailable in this situation. It is beneficial to conceal this frame by applying appropriate error concealment (EC) techniques. The case described above is explained in FIG.

Another specific frame, SID inhibited frame, is used instead of the regular SID frame when the first half of the TDMA frame containing the SID has been transmitted but the second half cannot be transmitted due to voice initiation. The pattern bits belonging to the second half of the SID frame to be transmitted are now inverted. This prevents the detection of the entire SID pattern at the receiver. The codec mode information bits indicating the codec mode indication remain the same from the original SID frame. The receiver will have a frame that is unavailable in the described situation, in which case it will continue to generate CN using the previous CN parameters. The receiver can also check the pattern transmitted in this particular case to detect the voice initiation with improved reliability. The case described above is described in FIG. 10.

According to the present invention, the SID frame is transmitted every n _FR frames (TCH / AFS) and every n _HR frames (TCH / AHS) during the inactive state. A suitable choice is n _FR = n _HR = 8. Phase aligned transmission and decoding of SID frames (alignment inferred from SACCH, as in current GCM systems) is one solution in today's GSM systems that facilitates SID frame decoding implementation. However, the proposed SID frame identification based on the total bit pattern provides high SID frame detection execution and enables a more flexible solution without a fixed aspect.

One example is to start the transmission of the SID frame in the third frame after the transmission of the SID_FIRST pattern, and then transmit the SID frame every eighth frame. Another solution is asynchronous SID transmission (ie not aligned to any fixed time structure). In one example, an SID frame is sent whenever a mode request is changed, with the constraint that perhaps a certain maximum of TDMA frames transmitted per 480 ms multiple frame has not yet been exceeded. Another improved solution is to send an SID frame if the CN parameters change meaningfully and some predetermined maximum of TDMA transmitted per 480 ms multiple frame has not yet been exceeded. Such a solution using asynchronous SID frame transmission may retreat to time aligned transmission whenever any predetermined minimum transmission requirement per time interval is not met.

Note that different bit patterns transmitted to identify different frame types may be partially corrupted by transmission errors. Correlation techniques may also be used to ensure reliable detection of patterns even in the face of channel errors. One possible solution is to calculate the number of matching bits when comparing the received bits with the pattern. As an example, if 70% of the bits match, then the receiver may consider the pattern found. Another solution using soft bit information is to accumulate received soft bits with a positive sign if the corresponding bit of the pattern is 1 and a negative sign if the corresponding bit is 0. This cumulative value can be normalized by the product of the pattern length and the maximum possible soft bit value. If the normalized value exceeds a certain threshold, for example 0.4, the receiver may consider the pattern found.

One other criterion that can be used for the SID frame is the CRC of the CN bit. If there is a CRC error, the frame is not considered a valid SID frame.

Considering the cost issue, it is desirable that the identification patterns do not require much memory to store them. For example, the identification pattern for SID_FIRST and the regular SID for TCH / AFS can be configured by repeating short 9-bit sequence seal ((228-16) / 9) = 24 times and then discarding the last 4 bits. have. Such a 9-bit sequence is, for example, {0,1,0,0,1,1,1,1,0}.

For THS / AHS, it is also important to avoid the possibility of decoding SID_FIRST frames as regular SID frames, and vice versa. Thus, identification patterns for SID and SID_FIRST are formed as distinct as possible.

For example, the pattern for the SID_FIRST frame may be the same as the pattern used in TCH / AFS. The pattern used in the regular SID frame may be configured by inverting the SID_FIRST pattern.

Also, the solution of transmitting only specific bit pattern and codec mode information in the SID_FIRST frame rather than transmitting CN parameters helps to keep the DTX efficiency at its maximum (ie, activity on the air interface is kept to a minimum). . At the same time, the detection reliability of the identification pattern can be maximum (except those used for transmission of codec mode information) since all available bits are used for the bit pattern. The problem with this, however, is that from the end of the voice until the first regular SID frame is received, the set of CN parameters for CN generation for that period is not obtained. The solution is to derive the CN parameters locally at the receiver by using the speech parameters of the last n frames before the speech ends. In general, encoders operate on hangovers. That is, even if the VAD detects voice inactivity, a certain number of m frames are still encoded in voice. Thus, the decoder can locally derive the CN parameters, for example by averaging the gain and spectral parameters of the remaining frame. That is, n = m. Another solution is to apply the last received set of CN parameters of the previous period of inactivity.

According to the invention, an AMR receiver incorporates a two-state model with active and inactive states. The purpose of this state model is to support the discrimination of voice / SID / non-transmitted frames. The transition from the active state to the inactive state requires the detection of the SID_FIRST frame following the speech frame.

The transition from inactivity to active state can be decoded without a CRC error and, optionally, a valid first speech frame, and voice initiation, for example representing a quality measure derived from a receiver / channel decoder that exceeds a certain threshold. It is necessary to detect the identification pattern. An example is an SFQ measurement (total bit error estimate) that should be below some threshold. The reliability of this state transition can be increased with the constraint that one or more frames must be able to decode without CRC errors and, optionally, without exceeding certain SFQ measurements. As described in FIG. 11, another criterion to help properly detect transitions from inactive to active states is that if block interleaving is used that requires less delay than diagonal interleaving of speech frames for SID frames, the SID is used. The received frame immediately following the frame can never be a voice frame. 12 illustrates this criterion for an example of TCH / AHS.

Another way to improve the detection of the first speech frames and to help distinguish them from untransmitted frames is to access measurements from other components of the receiver (eg, RF receiver or equalizer). Examples of such measurements are derived measurements such as half-wave and interference strength estimates and C / I ratios.

Another way to improve both SID_FIRST and first speech frame identification execution is to transmit TDMA frames containing them with increased transmit power.

According to the invention, the following solutions are suitable for defining the codec mode for speech initiation after a period of inactivity:

(a) Choice of the most robust codec mode, or the nth robust codec mode. The safest solution is to choose n = 1. Codec mode indications do not need to be sent. The problem with n = 1 is that too robust codec mode with low inherent sound quality is selected if the channel is good.

(b) Selection of the same codec mode as the currently active link. This is caused by the fact that the uplink and downlink channel quality are similar. The sender of the link that resumes voice transmission uses the codec mode requested by the incoming active link. The receiving side of the link being reactivated recognizes the codec mode used because it is identical to the codec mode request it is receiving for application to the outgoing current active link. The scheme may be formed to be more robust if a mode for voice initiation is selected that is n more robust (if such a robust mode exists) than the mode of the current activity link.

(c) Selection of the same codec mode that was selected at the end of the last speech period before the inactivity period. This is caused by the fact that radio channel conditions do not generally change very quickly. The scheme may be formed to be more robust if a mode for voice initiation is selected that is n (eg n = 1) mode that is more robust (if such a more robust mode exists) than was used at the end of the final speech period. Can be.

(d) Selection based on the measurement of inactive links. Since the transmission of the inactive link is not completely stopped, link quality measurement is possible. Correspondence measurement reports or codec mode requests / commands are sent over the activity link. When the inactive link resumes voice transmission, the codec mode corresponding to the last received codec mode request is selected.

Advantageously, the solutions (a), (b) and (c) above can take advantage of the fact that a codec mode request for an inactive link does not need to be sent. The active link can thus reduce the transmission capacity for codec mode requirements and use it for some other purpose. One example is to use this transport capacity to more securely protect the transmission of codec mode indications.

In addition to the above techniques for implementing DTX in AMR systems, the present invention further provides techniques for implementing rapid configuration changes in AMR systems. The purpose of these techniques is to enable rapid configuration changes that cannot be made using current low speed control channels. In addition, current control channels cannot guarantee that configuration changes are synchronized with voice data transmission. Like the DTX mechanism, the configuration change mechanism is based on in-band signaling. Applications are, for example, a change in active codec set and a phase change in codec mode information with respect to random free operation (TFO) (to minimize transmission delay). Further general applications will be to move to one of the current GCM codecs (FR, EFR, HR), or to apply in the future, for example, wideband codecs, voice and data, or multiple media. Like the DTX mechanism, the configuration change mechanism is described with respect to TCH / AFS and TCH / AHS in GSM systems, but is applicable in other situations as well.

The configuration change mechanism is based on frame stealing similar to the well known FACCH frame stealing (ie, voice frames are replaced with configuration change frames) and is referred to as extended signaling in the duplex. Since the extended signaling mechanism is only used during occasional connections and only a few voice frames will be stealed, the error concealment unit at the receiver may make it impossible to actually hear the frame stealing.

According to the present invention, the extended frame is in a format similar to the above-described SID frame. They are identified at the total bit level by specific identification patterns. Like SID frames, they contain this pattern and one or two message fields. One field contains the actual channel encoded extended message and the other field contains the codec mode information. As an example, the extended frame may include 456 bits and may be in exactly the same frame format as the SID frame (for example, see FIG. 4) in which the CN field is replaced by the extended message.

The payload to be sent by the extension mechanism is called an extension message. An extension message consists of a number of last bits that can be grouped into logical units. For example, according to the present invention, GSM 05.09: digital cellular telecommunication system (phase 2+); See Link Adaptation (Draft ETSIEN 301 709 V7.1.0 (1999-11), the entire contents of which are incorporated herein by reference.

The extension message may be a channel encoded with any suitable channel coding scheme, such as, for example, block or convolutional coding. One solution to cost effectiveness is to use channel coding exactly as used in the CN parameter in the SID frame as described above. This follows that when following the above example solution with 35 CN bits, an extended message of 35 net bits is protected with a 14 bit CRC and then convolutionally with 1/4 code rate and constraint length k = 5. It is encoded.

As in the case of the SID frame, the codec mode information field may include both a codec mode indication and a codec mode command / request encoded in a block or convolutional manner.

The extended frames, like voice frames, are interleaved block diagonally. This means that when taking the example solution with an extended frame of 456 total bits, the extended frame replaces one speech frame in TCH / AFS and two speech frames in TCH / AHS.

In the case of TCH / AHS, this is not necessarily two consecutive frames, but this takes this in the above example solution. Not stealing two consecutive frames is beneficial for error concealment to hide stealing. On the other hand, stealing two consecutive voice frames is beneficial in terms of delay in transmission of the extended message. Interleaving is performed in such a way that the first half of the extended frame (228 bits, see FIG. 4) replaces the first speech frame. It is important that this first half contains an extended identification pattern. This allows the receiver to check this pattern. After finding the pattern, the receiver can place a second concealed speech frame that includes the second half of the extended frame.

In order not to interfere with the regular transmission of codec mode information, the interleaver may further map one of the codec mode information code words to the bit position of the first concealed speech frame. As a result, another codec mode information code word is mapped to the bit position of the second concealed speech frame. In addition, placing codec mode information, i.e., codec mode indication and request / command into the codec mode field, is executed with respect to the codec mode information phase during the transmission of regular speech frames. For example, if the first half of the extended frame replaces a voice frame that would have included a codec mode indication, then the first half of the extended frame should still send the codec mode indication.

Note that the extension mechanism described above can also be used in conjunction with the DTX mechanism described above. Thus, according to the present invention, the extended frame may replace not only the speech frame but also all other types of frames, namely SID_FIRST, regular SID, NoTX, and speech start frame. Considering the case where an extended frame should be transmitted during the period of inactivity, applying block interleaving is effective in terms of transmission resource usage, as is performed for SID frames. However, since the extension mechanism is only intended to be used occasionally, transmission resource usage is not the most important criterion. Rather, it is important to make the cost of implementation efficient and to reduce complexity. Thus, it is an advantageous solution to maintain block diagonal interleaving, channel coding, and frame format, which are also used in extended frames during speech.

Using block diagonal interleaving for extended frames during DTX means that there are half bursts that are not defined by interleaving. For TCH / AFS, the even positions of the last four bursts and the odd positions of the first four busters containing the extended frame are not determined. The undetermined bit is not a problem in itself, but the following problem can be solved by setting the undetermined positions appropriately. Consider the case of voice initiation. As described above, the speech initiation frame is marked with an initiation pattern to better identify the frame as an initiation and to better identify the codec mode used for the initiation speech frame. If extended frames should be transmitted at the same time, this will replace the start frame. Thus, in subsequent speech frames, since the starting pattern has been stealed, it is more difficult to identify them as speech frames.

According to the present invention, with or without a disclosure, this problem is solved by filling the first half of the undetermined bit (odd positions) with the initiation pattern. In fact, if there is no initiation, it is necessary to signal that the inactive state continues. This is done by sending SID_FIRST immediately following the extended frame. This defines the second half of the other unused bit (even positions). This solution is more beneficial in terms of implementation costs. This allows processing of extended frames apart from channel coding, exactly as if it were voice. 13 and 14 describe the above solutions for TCH / AFS and TCH / AHS, respectively.

delete

Note that a voice frame that has been stealed for expansion purposes cannot be rescheduled for transmission after expansion, because it can increase the voice transmission delay. However, SID frames affected by the extended frame transmission may be rescheduled for transmission immediately after the extended frame transmission. Preferably, this helps to maintain high subjective comfort noise signal quality. An example solution is provided in GSM 06.93 cited above.

In order to ensure the correct reception of extended messages and to set up appropriate routines for error incidents, extended protocols are proposed. Example solutions are provided in GSM 05.09 cited above.

Those skilled in the art will recognize that the present invention is not limited to the specific illustrated embodiments described herein for the purpose of description, and that many other embodiments are also contemplated. Therefore, the scope of the present invention is defined by the claims accompanying it rather than the foregoing description, and all equivalents consistent with the meaning of the claims are to be interpreted as included therein.

Claims (28)

A method for performing discrete transmission (DTX) in an adaptive multirate (AMR) communication system in which source data is channel encoded and interleaved for transmission from a first component in a system to a second component in the system. The method is as follows: Detecting a period of source data inactivity by the first component; And Transmitting silence descriptor (SID) frames containing codec mode information from the first component to the second component during the source data inactivity state period, Some of the transmitted SID frames are interleaved using a different interleaving algorithm when compared to that used for the channel encoded source data. The method of claim 1, wherein the source data is block diagonal interleaved and some of the SID frames are block interleaved. The method of claim 1, wherein the SID frames also include comfort noise (CN) parameters. The method of claim 1, wherein transmitting the SID frames comprises: Transmitting a SID frame of a first type indicating a transition from a source data active state to a source data inactive state; Periodically transmitting a second type of SID frame during a source data inactivity state; And Transmitting a third type SID frame indicating a transition from a source data inactive state to a source data active state; How to include. delete The method of claim 1, wherein each of the SID frames includes a bit pattern for identifying the SID frame as a particular type of SID frame. The method of claim 6, wherein the bit patterns are a gross bit pattern. 2. The method of claim 1 wherein the source data is voice and the communication system is one of a time division multiple access (TDMA) wireless system, a frequency division multiple access (FDMA) wireless system, and a code division multiple access (CDMA) wireless system. The method of claim 1, wherein extended frames are transmitted to cause a configuration change, wherein the extended frames can replace source data frames, SID frames, or non-transmitted (NoTX) frames. 10. The method of claim 9, wherein the SID frames are block interleaved and the extended frames are block diagonally interleaved. 10. The method of claim 9, wherein the extended frame is used to change a codec mode set. 10. The method of claim 9, wherein the extended frame is used to change the phase of codec information. 2. The method of claim 1 wherein active speech source data is block diagonally interleaved, unused bits within an interleaving scheme for a final speech frame are used for a particular bit pattern that indicates the end of speech, and the interleaving for a first frame. Unused bits in a scheme are used for a specific bit pattern that marks the beginning of speech. delete delete delete delete delete delete delete delete delete delete delete Adaptive Multirate (AMR) voice communication system, A first component for transmitting channel encoded and interleaved speech data frames as a second component, The first component detects a period of speech inactivity and transmits interleaved silence descriptor (SID) frames instead of the channel encoded and interleaved speech data frames during the period of speech inactivity-the silence descriptor frames Contains codec mode information-, At least some of the SID frames are interleaved using a different interleaving algorithm as compared to that used for the channel encoded speech data frame. 27. The system of claim 25, wherein the channel encoded speech data frames are block diagonal interleaved and some of the SID frames are block interleaved. delete delete

Images