HK1064524B - Testing loops for channel codecs - Google Patents

Description

Test loop for channel codec

Technical Field

The present invention relates to a method for measuring decoding performance in a telecommunication system.

Background

In wireless digital telecommunications, analog speech information must be encoded into digital form prior to transmission and then secured using channel coding to ensure adequate voice quality upon reception of the signal. For example, in conventional GSM speech coding, the speech codec has a fixed rate. In the GSM system there are two full rate speech codecs and one half rate speech codec in use. Full-rate speech codecs have an output bit rate of 13 or 12.2 kbits/sec, while half-rate speech codecs provide an output bit rate of 5.6 kbits/sec. These output bits representing the encoded speech parameters are fed to a channel encoder. Channel coding is a set of functions responsible for adding redundancy to an information sequence. Encoding is typically performed on a fixed number of input bits. The output bit rate of the channel encoder is adjusted to 22.8 kbits/sec in a full rate traffic channel or to 11.4 kbits/sec in a half rate traffic channel.

Thus, all conventional GSM codecs operate with a fixed division between speech and channel coding bit rates regardless of the quality of the channel. These bit rates never change unless a traffic channel change occurs, which is a slow process. This solution, which is rather inflexible both in terms of desired speech quality and system capacity optimization, has therefore led to the development of AMR codecs (adaptive multi-rate).

The AMR codec adapts the division between speech and channel coding bit rate according to the channel quality in order to provide the best possible overall speech quality. AMR vocoders include multi-rate vocoders, source controlled rate schemes including voice activity detectors and comfort noise generation systems, and error concealment mechanisms to combat the effects of transmission errors and lost packets. The multi-rate speech coder is a single integrated speech codec with eight source rates from 4.75 kbit/s to 12.2 kbit/s and a low-rate background noise coding mode.

For example, in the GSM system, there are several performance criteria set for the codec used, and the performance can be measured by, for example, the Frame Erasure Rate (FER), Bit Error Rate (BER) or Residual Bit Error Rate (RBER) of the received data on any traffic channel TCH. Furthermore, to enable automatic measurement of performance, a set of test loops have been developed. A set of predefined test loops is implemented into a mobile station connected to a system simulator. The system simulator activates a specific test loop and starts providing random or predefined test data into the codec. The mobile station sends the obtained data back to the system emulator after performing channel decoding. The system simulator can then compare the data being looped back with the data being transmitted. In this way, the performance of the channel decoder part, e.g. a codec, can be measured with respect to several criteria.

A problem associated with the above-described solution is that these test loops are designed to be particularly suitable for the previous GSM codec. However, the AMR codec includes features not available from the previous codec, and therefore all the features of the AMR codec cannot be tested using the known test loop.

Disclosure of Invention

It is an object of the present invention to provide an improved method and apparatus for carrying out the method which avoids at least some of the problems described above. The object is achieved with a method and an apparatus which are characterized by what is stated in the independent claims. Preferred embodiments of the invention are disclosed in the dependent claims.

The idea on which the invention is based is: when determining decoding performance in a telecommunication system comprising a decoder and test equipment for supplying test data to the decoder, measurements are initiated by generating in the test equipment a test data comprising speech parameters and in-band data fields which are channel encoded in a frame format, preferably a speech frame format, and then transmitted to the decoder for decoding. The decoder extracts at least a portion of the in-band data field from the decoded test data and sends the at least a portion of the in-band data field back to the test device so that no speech parameters or any other data is sent. Decoding performance is then determined by comparing the transmitted in-band data field with the received in-band data field in the test equipment.

One advantage of the method and device according to the invention is that the performance of the inband decoder can also be measured. Another advantage of the present invention is that implementation problems related to different uplink and downlink speech codec bit rates are reduced because only in-band data is looped back from the decoder. Another advantage of the present invention is that existing test equipment can be used with only minor modifications.

Drawings

The invention is described in more detail below, in connection with preferred embodiments and with reference to the accompanying drawings, in which:

figure 1 shows a radio system using the method of the invention;

fig. 2 shows the general structure of a channel coding chain in an encoder;

FIG. 3 shows the composition of TCH/AFS frames for different codec modes;

FIG. 4 shows the composition of TCH/AHS frames for different codec modes;

FIG. 5 is a flow chart illustrating a new testing method according to the present invention; and

fig. 6 is a block diagram illustrating a test apparatus implementing a method according to the present invention.

Detailed Description

The invention is described in more detail below using the GSM system as a preferred platform for embodiments of the invention. However, the invention is not limited to GSM systems only, but it can be used in any corresponding system where the implementation of the test loop encounters similar problems. Therefore, the present invention can be applied to, for example, a WCDMA (wideband code division multiple access) system in which an AMR (adaptive multi-rate) codec is also supported.

Fig. 1 shows an example of a radio system, some parts of which use the method of the invention. The presented cellular radio system comprises a base station controller 120, a base transceiver station 110 and a set of user terminals 100, 101. The base transceiver stations 110 and the user terminals act as transceivers in a cellular radio system. The user terminals establish connections with each other using signals propagating through the base transceiver stations 110. The user terminal 100 may be, for example, a mobile phone. The radio system presented in fig. 1 may be, for example, a GSM system and, for example, a TDMA multiple access method may be used in the radio system.

In the GSM system, several logical channels are transmitted on a grid (grid) of physical channels. Each logical channel performs a specific task. Logical channels can be divided into two categories: traffic Channels (TCH) and Control Channels (CCH). GSM voice traffic channels are TCH/FS (full rate voice channel), TCH/HS (half rate voice channel), TCH/EFS (EFR voice channel), TCH/AFS (AMR voice channel on FR) and TCH/AHS (AMR voice channel on HR). In addition, several control channels are defined in GSM, most of which are used for call setup and for synchronization. However, when an AMR call is active, SACCH (slow associated control channel), FACCH (fast associated control channel) and RATSCCH (robust AMR traffic synchronized control channel) channels are involved. Both SACCH and FACCH are used for the transmission of signalling data during a connection, but one SACCH slot is allocated in every 26 th TDMA frame, while the FACCH channel is used only when needed. Also RATSCCH, which is used to modify the AMR configuration over the radio interface during a connection, is used only when needed. When a FACCH or RATSCCH is needed, the necessary time slots are allocated for FACCH or RATSCCH by "stealing" them from TCH speech frames.

In conventional GSM speech coding, the speech codec has a fixed rate. In the GSM system there are three speech codecs in use: full Rate (FR) speech codecs based on the RPE-LTP method (regular pulse excitation-long term prediction), Half Rate (HR) speech codecs based on the CELP/VCELP method (codebook excited linear prediction) and Enhanced Full Rate (EFR) speech codecs based on the ACELP method (algebraic codebook excited linear prediction). The speech codec provides the speech parameters to the channel codec every 20 milliseconds. Since the active call logical channel mapping lasts 120 ms, it contains 6 speech frames. In full rate traffic channels (TCH/FS) and full rate traffic channels using enhanced coding (TCH/EFS), a new speech frame is sent every 4 th burst containing TCH information. For each 20 ms speech frame the full rate speech codec FR provides 260 bits and the enhanced full rate speech codec EFR provides 244 bits representing the encoded speech parameters, resulting in an output bit rate of 13 kbit/s and 12.2 kbit/s, respectively. In a half rate traffic channel (TCH/HS), a new speech frame is sent every 2 nd burst containing TCH information. For each 20 ms speech frame, the half-rate speech codec HR provides 112 bits representing the encoded speech parameters, resulting in an output bit rate of 5.6 kbit/s.

These output bits representing the encoded speech parameters are fed into a channel encoder. Channel coding is a set of functions responsible for adding redundancy to an information sequence. Encoding is typically performed on a fixed number of input bits. Higher coding gain is achieved by increasing the complexity of the coding. However, transmission delays and limited hardware resource limitations can be used for complexity in a real-time environment.

Referring now to fig. 2, a channel coding chain in an encoder is shown. The channel coding of speech parameters consists of several blocks. Bit reordering (200) is performed on bits of the speech parameters according to subjective importance, thereby dividing the bits into classes 1A, 1B, and 2. For the most important bits, i.e. the type 1A bits, a CRC is calculated (cyclic redundancy check, 202). The CRC technique transmits few additional bits that are used by the receiver to error the transmitted frame. Type 1B bits are not protected by CRC. Type 1A and 1B bits are protected by convolutional coding (204), which is a method of adding redundancy to bits transmitted in a channel. The convolutional encoder produces more output bits than input bits. The manner in which redundancy is added allows the receiver to perform a maximum likelihood algorithm on the convolutionally encoded bits to allow correction of signal errors generated in the transmission. The number of bits that can be transmitted in the channel is limited. Puncturing (206) is a method of reducing the number of bits transmitted over a channel by removing bits from convolutionally encoded data. The decoder knows which bits are punctured and adds placeholders to them. In the FR channel, 456 bits are transmitted every 20 milliseconds, resulting in a total rate of 22.8 kbit/sec in the full rate traffic channel. In the HR channel, 228 bits can be sent every 20 milliseconds, respectively, resulting in a total rate of 11.4 kbit/sec, which is exactly half the total rate used in the full rate traffic channel.

As described above, all previous GSM codecs operate with a fixed division between speech and channel coding bit rates regardless of channel quality. These bit rates never change unless a traffic channel change occurs (from FR to HR or vice versa), which is also a slow process requiring layer 3(L3) signaling. The fixed partitioning does not use the fact that: the protection provided by channel coding is highly dependent on the channel conditions. When the channel conditions are good, a lower channel coding bit rate can be used, allowing a higher bit rate for the speech codec. Thus, allowing a dynamic split between speech and channel coding bit rates improves overall speech quality. The formation of this idea brings about the standardization of AMR codecs.

The AMR codec adapts the error protection level to the radio channel and traffic conditions so that it always aims at selecting the best channel and codec mode (speech and channel bit rate) to achieve the best overall speech quality. The AMR codec operates in the GSM FR or HR channel and it also provides the user with a speech quality comparable to wireline for half rate channels in good channel conditions.

AMR vocoders include multi-rate vocoders, source controlled rate schemes containing voice activity detectors and comfort noise generation systems, and error concealment mechanisms that combat the effects of transmission errors and lost packets. The multi-rate speech coder is a single integrated speech codec with eight source rates from 4.75 kbit/s to 12.2 kbit/s and a low-rate background noise coding mode. The speech encoder can switch its bit rate every 20 ms speech frame on command.

The AMR codec contains eight speech codecs with bit rates of 12.2, 10.2, 7.95, 7.4, 6.7, 5.9, 5.15 and 4.75 kbits/sec. All speech codecs are defined for full rate channels and the six lowest speech codecs are defined for half rate channels, as shown in the following table.

	12.2	10.2	7.95	7.4	6.7	5.9	5.15	4.75
									TCH/AFS	×	×	×	×	×	×	×	×
TCH/AHS			×	×	×	×	×	×

One mobile station must implement all codec modes. However, the network can support any combination thereof. For AMR, a codec mode is selected from a group of codec modes (ACS, active codec set) comprising 1-4 AMR codec modes. The group may be reconfigured during the call setup phase, in a handover situation, or using RATSCCH signaling. Each codec mode provides a different level of error protection by different allocations between speech and channel coding. All speech codec modes are allowed to change without L3 signaling intervention, enabling fast switching between modes when channel conditions change.

Fig. 3 illustrates the formation of TCH/AFS frames for different codec modes. The frame is constructed starting with 244 bits output by the speech codec, using the 12.2 kbit/sec case. The speech frame bits are reordered and divided into type 1A (81 bits) and 1B (163 bits). For the protection of 81 type 1A bits, a 6 bit CRC is calculated. 4 tail bits are added to the 250-bit block, which are used for termination of the channel encoder. An 1/2 rate convolutional code (244+6+4) is performed on the 254 bit block resulting in a 508 bit block. Then, the block of 508 bits is punctured, thereby reducing the number of bits to 448 bits. Finally, 8 bits containing inband data are added. The final block is 456 bits long.

As shown in fig. 3, all TCH/AFS channel coded frames have the same length (456 bits), even though the number of input bits (speech parameters) is different for different modes. By varying the convolutional coding rate as well as the puncturing rate for each mode, a different number of input bits are coded into exactly 456 output bits. The 456 bits sent every 20 milliseconds, which brings a total rate of 22.8 kbit/sec, utilizes all the bits available from the full rate traffic channel of the GSM system.

Fig. 4 shows the formation of TCH/AHS frames for six different codec modes, respectively. The principle of frame construction is similar to the case of TCH/AFS frames, with a few exceptions. In bit reordering, bits are divided into type 1A, 1B and 2 bits, whereas in TCH/AFS frames only type 1A and 1B are used. These type 2 bits are not convolutionally encoded. In addition, only 4 inband data bits are added to the convolutionally encoded frame. In all TCH/AHS codec modes, the channel coded frame is 228 bits long. 228 bits sent every 20 milliseconds, which brings a total rate of 11.4 kbit/sec, satisfies the GSM system's requirements for half rate traffic channels.

As mentioned earlier, 8 speech codec modes are defined for AMR and AMR codecs can be used on existing FR and HR channels. Thus, 14 different codec modes are defined for AMR (8 for TCH/AFS channels and 6 for TCH/AHS channels).

The link adaptation process is responsible for measuring the channel quality. The best speech and channel codec is selected by mode adaptation based on the quality and possibly network constraints (e.g. network load). Both the Mobile Station (MS) and the Base Transceiver Station (BTS) perform channel quality estimation for their own reception paths. Based on the channel quality measurements, the BTS sends a codec mode command (CMC, mode used by the MS in the uplink) to the MS, and the MS sends a codec mode request (CMR, requested for mode in the downlink) to the BTS. This signaling is sent in-band with the voice data. The codec mode in the uplink may be different from the codec mode used in the downlink, but the channel mode (full rate or half rate) must be the same. In-band signaling must be designed to allow fast adaptation to fast channel changes.

The network controls the uplink and downlink codec modes and the channel mode. The mobile station must obey the codec mode command from the network and the network can use any supplemental information to determine the downlink and uplink codec modes.

In the GSM system, for example, the channel coding algorithm is specified in detail. Instead of specifying the channel decoder algorithm, performance criteria are defined and must be met by the MS. Several performance criteria are set for the channel codec used in the GSM system, which performance can be measured by, for example, the Frame Erasure Rate (FER), Bit Error Rate (BER) or Residual Bit Error Rate (RBER) of data received on any traffic channel TCH. For GSM systems, in the document "3 GPP TS05.05 V8.7.1, Digital cellular telecommunications system (phase 2 +); the Radio transmission and reception "defines the standard more precisely. To facilitate the development and implementation of channel codecs and to measure the performance of the receiver, a special device called a System Simulator (SS) is defined for e.g. type approval. A set of test loops have been developed to measure the performance of a channel decoder. A predefined test loop is activated in a mobile station connected to a system simulator and performance is measured with respect to several criteria. For the GSM system, in the document "GSM 04.14 ETSI TS 101293 v8.1.0, digital cellular telecommunications system (phase 2 +); industrial demand type requirements and interworking; special compliance testing function) defines these test loops more accurately.

These test loops are designed to be particularly suitable for the previous GSM codec. However, the AMR codec comprises features not involved in previous codecs, so that all the features of the AMR codec cannot be tested with the known test loop. The present invention addresses at least some of the problems involved in AMR testing.

One problem relates to determining in-band signaling decoding performance. As shown in fig. 3 and 4, an AMR encoded traffic channel frame always includes some control bits sent along with the speech bits. These bits are called inband signaling bits. These bits are used to enable a change of codec mode without any further signalling frames. Since there are a maximum of four modes in a mode group, only two bits are required to encode the inband information. To aid decoding in difficult channel conditions, the two bits are mapped to a longer bit pattern: 8 bits on TCH/AFS and 4 bits on TCH/AHS.

The information transmitted in-band depends on the direction. In the downlink direction (from BTS to MS), two different pieces of information are time multiplexed into two consecutive speech frames. In the first frame, a mode command MC is sent from the BTS to the MS, whereby the BTS commands the mode that the MS must use in the uplink. In the second frame, a mode indication MI is transmitted from the BTS to the MS, so that the BTS informs the MS of the mode it uses in the downlink. Also in the uplink direction (from MS to BTS), two different pieces of information are time multiplexed into two consecutive speech frames. In the first frame, a mode request MR is transmitted from the MS to the BTS, so that the MS requests the BTS to use a certain mode in the downlink. In the second frame, a mode indication MI is sent from the MS to the BTS, so that the MS informs the BTS of the mode it uses in the uplink. The information transmitted inband is always time multiplexed, i.e. every other frame contains the current pattern and every other frame contains the commanded/requested pattern.

When a 20 MS frame is received by the MS, it is processed by the channel decoder. The output of the channel codec is the channel decoded speech parameters together with the information transmitted in-band. If the information is a Mode Command (MC), the MS will modify the voice mode it uses in the uplink according to the command, since the MS must always obey the commanded Mode (MC) from the BTS. This used uplink mode will be signaled to the BTS via an uplink mode indication transmitted in-band.

Since the preceding traffic channel frame of the fixed rate channel codec does not include any inband data, there is no existing test method to measure the performance of the inband decoder in all cases. If one tries to measure the in-band decoder performance with the current test loop and test equipment (system simulator SS), the MS will follow the received Mode Command (MC) and change its uplink Mode Indication (MI) accordingly. The tester SS may then compare the received MI against the previously transmitted MC. If the two are similar, the inband decoder can be considered to be working correctly. If they are different, it informs the MS that the MC from the BTS was not decoded correctly. Based on these checks, the SS can calculate the performance of the inband decoder.

A problem arises when trying to assess the performance of an MI inband decoder. The downlink MI has no direct impact on any uplink inband signaled information. As mentioned above, the uplink MI is directly affected by the downlink MC. Mode Request (MR) is maintained with respect to retaining two time multiplexed inband information. The mode request is generated using a mobile station uplink adaptation algorithm and is not modified directly by the downlink MI. For this reason, the SS cannot calculate the performance of the MI inband decoder.

According to an erroneous decoding of the downlink MI, which follows the erroneous decoding of the speech parameters, the CRC check fails, and then the frame is declared bad. If the previous test loop is activated, the speech parameters decoded in error are sent back to the tester SS. The SS may compare the transmitted speech parameters with the looped-back speech parameters in order to determine the performance of the MI in-band decoder. However, the channel coding of the inband bits is much stronger than the channel coding of the speech parameters, and therefore the decoding of the speech parameters is more likely to fail than the decoding of the inband parameters. Thus, the measured performance will be one of the speech parameter decoders and not one of the inband decoders.

A new internal test loop has been developed to overcome this problem. In the new test loop, the link adaptation algorithm is bypassed and replaced by a function that loops back the received in-band data. This is done regardless of the in-band signaling phase. This leads to two possible situations: the received MC can be sent in the uplink as an MI, and then the received MI is looped back as an MR. In other possible cases, the received MC can be sent in the uplink as an MR and the received MI looped back as an MI. Since the goal of the loop is to compute the in-band decoding performance, the speech parameters sent by the SS are not looped back from the MS, but they are encoded as zeros. Advantageously, this reduces implementation problems related to different uplink and downlink speech codec bit rates. Only the inband signaling mode, i.e. only the inband bits, no speech parameters are sent back to the SS and the performance of the inband decoder can advantageously be measured. Depending on the received in-band signaling pattern, the frame error rate (TCH/AxS-INB FER) for the in-band channel can be determined, for example.

The method according to the new test loop is described with reference to the flow chart in fig. 5. To establish a transparent test loop for TCH frames, a TCH must be active between the SS and the MS. The TCH can be AMR speech on full rate channels or half rate channels of any rate specified in the GSM system. The test LOOP is activated in the MS by sending a correct command message to the MS, which may be, for example, a CLOSE _ TCH _ LOOP _ CMD message according to the GSM system. The SS commands the MS to CLOSE its TCP LOOP (500) by sending a CLOSE _ TCH _ LOOP _ CMD message that specifies that the TCH is looped and that the decoded in-band signaling information is looped back by the MS. The SS then starts a timer TT01(502), which sets a corresponding time limit for the MS. If no TCH is active, or any test LOOP has been closed (504), the MS will ignore any CLOSE _ TCH _ LOOP _ CMD message (506). If a TCH is active, the MS will CLOSE its TCH LOOP for the specified TCH and send a CLOSE _ TCH _ LOOP _ ACK back to the SS (508). Upon receiving the message, the SS stops the timer TT01 (510).

After the MS closes its TCH loop, each in-band signal decision is taken from the output of the channel decoder (512) and input to the channel encoder (514). By setting the input frame to the channel encoder to zero (516), the transmitted speech parameters are not looped back. The in-band signal decisions input to the channel encoder are sent to the SS on the same TCH uplink (518). This is advantageously done regardless of link adaptation so that the decoded inband information is looped back directly to the SS. The SS measures the performance of the in-band decoder according to the received in-band signaling pattern, e.g. by determining the frame error rate (TCH/AxS-INBFER) for the in-band channel (520).

The content of the CLOSE _ TCH _ LOOP _ CMD message is more accurately defined in the above-mentioned document GSM 04.14. This message is only sent in the SS to MS direction. The CLOSE _ TCH _ LOOP _ CMD message includes four information elements: a protocol discriminator field and a hop indicator field, both of which have a length of four bits and are more accurately defined in the documents "GSM 04.07, v.7.3.0, sect.11.1.1 and 11.1.2", a message type field having a length of eight bits, all defined as zero, and a subchannel field also having a length of eight bits. Of the subchannel field bits, five bits have special meaning to define the message content, and they are referred to as X, Y, Z, A and B bits. Three bits are spare bits set to zero.

The activation of the test LOOP according to the invention can be implemented with a CLOSE _ TCH _ LOOP _ CMD message if one of the spare bits is also advantageously assigned a special meaning that defines the message content. This new bit may be referred to as, for example, the C bit. Then, defining the C bit to have a value of one, a new message content may be defined with a particular bit combination. For example, the following bit combinations may be defined: a-1, B-0 and C-1, means that if the TCH being looped is TCH/AxS, the decoded in-band signaling information is looped back. The value of X bits indicates whether only one full-rate channel is active or which of the possibly available sub-channels is used. The values of the Y and Z bits may be discarded.

According to a second embodiment of the present invention, a test sequence of an in-band data mode to be used by the SS is provided to the MS. This provision may be performed before the test loop is activated or during the test set-up. The SS activates the test LOOP in the MS, for example by sending a CLOSE _ TCH _ LOOP _ CMD message, and starts sending the test sequence. In the MS, a counter is implemented, which is incremented each time the decoded inband data does not fit the expected result. When the test sequence is completely looped back, the value of the counter can be checked from the MS or it can be sent to the SS, from which the performance of the inband decoder can be derived.

According to a third embodiment of the invention, the link adaptation algorithm is kept in an active state and the MS follows the mode command MC sent by the SS. Then, only the mode indication MI according to the commanded mode MC is sent back to the SS. The speech parameters sent by the SS are not looped back from the MS, but they are encoded as zeros. The SS compares the received mode indication MI with the transmitted mode command MC and if they coincide, the mode command MC decoding can advantageously be measured. However, since only every second frame is tested for SS, a separate test loop must be utilized to measure the performance of the mode indication MI decoding.

The block diagram in fig. 6 represents a device that can be applied to a test arrangement according to the invention. The system simulator 600 includes a generator 602 for generating random/fixed speech parameter patterns, which are then input to a channel encoder 604 for encoding. The channel coded speech frames are then provided to the transmitting means 606 for further transmission via the channel emulator 608 to the mobile station 610. The mobile station 610 comprises a receiving means 612 for receiving said transmission, from which channel encoded speech frames are input to a channel decoder 614. The mobile station 610 comprises means 616 for implementing test loops and for executing a particular test loop according to the instructions given by the system simulator 600. As described above, the test LOOP to be used may be defined, for example, using a CLOSE _ TCH _ LOOP _ CDM message. The output of the test loop is provided to a channel encoder 618 for encoding. The channel coded data is then provided to the transmitting means 620 for further transmission to the system simulator 600. The system simulator 600 further comprises receiving means 622 for receiving said transmission, from which channel encoded data is input to a channel decoder 624. The system simulator 600 comprises comparing means 626 for comparing the received data with the transmitted pattern, and as a result of said comparison the performance of the decoding can be determined.

It is obvious to a person skilled in the art that with the advancement of technology, the invention may be implemented in various ways. The invention and its embodiments are thus not limited by the examples described above but may vary within the claims presented below.

Claims (12)

1. In a telecommunications system including a decoder and test equipment for providing test data to the decoder, a method of determining decoding performance, the method comprising the steps of:

test data including parameters of channel coding and inband data is generated,

a traffic channel of the telecommunications system is activated,

sending test data from the test equipment to a decoder for decoding, characterized by:

extracting at least a portion of the inband data from the decoded test data,

one link adaptation process of the decoder is bypassed,

sending at least a portion of the in-band data back to the test equipment, an

The decoding performance is determined by comparing the transmitted inband data with the received inband data in the test equipment.

2. The method of claim 1, wherein:

test data is sent from the test device to the decoder in a downlink traffic channel and from the decoder to the test device in an uplink traffic channel.

3. The method of claim 2, wherein:

in-band data is sent back to the test equipment in the first available uplink traffic channel time frame.

4. A method according to claim 2 or 3, characterized in that:

sending a message from the test equipment to activate a test loop in the decoder, the test loop being functionally implemented in connection with the decoder, before sending the test data, an

The message is acknowledged from the decoder to the test equipment in response to the traffic channel being activated.

5. The method of claim 4, wherein:

the message is a bit combination of the CLOSE _ TCH _ LOOP _ CMD message according to the GSM system.

6. A method according to any one of claims 1-3, characterized in that:

the parameters of the channel coding are speech parameters.

7. A method according to any one of claims 1-3, characterized in that:

channel decoding performance of a Mode Indication (MI) inband data field is determined in an AMR full rate or half rate speech channel.

8. A test device for determining the performance of a decoder, the test device being arranged to be functionally connected to the decoder, the test device comprising:

composing means for composing test data comprising parameters of channel coding and inband data,

means for activating a traffic channel to a decoder,

a transmitter for transmitting test data to a decoder for decoding, further comprising:

control means for sending a command to the decoder to bypass its link adaptation process,

a receiver for receiving at least a portion of the inband data, an

A comparator for determining decoding performance by comparing the transmitted inband data with the received inband data.

9. Test device according to claim 8, characterized in that the test device is arranged to:

sending test data to a decoder in a downlink traffic channel, an

Test data is received from a decoder in an uplink traffic channel.

10. Test device according to claim 9, characterized in that the test device is arranged to:

sending a message to the decoder to activate a test loop in the decoder, the test loop being functionally implemented in connection with the decoder, before sending the test data, an

Receiving an acknowledgement of the message from a decoder in response to a traffic channel being activated.

11. A mobile station, comprising:

a receiver for receiving test data including channel coded parameters and in-band data from a test device,

a decoder for decoding test data, further comprising:

extracting means for extracting at least a part of the inband data from the decoded test data,

control means for controlling a link adaptation process of a decoder to be bypassed, and

a transmitter for transmitting at least a portion of the in-band data back to the test equipment.

12. The mobile station of claim 11, wherein:

the in-band data is arranged to be transmitted back to the test equipment in the first available uplink traffic channel time frame.