You are on page 1of 7

UMTS

Javier Sanchez & Mamadou Tnioune


Copyright 02007, ISTE Ltd.

Appendix 1

AMR Codec in UMTS

Originally developed to be used in GSM by the ETSI, the Adaptive Multi-Rate


(AMR) speech codec [TS 26.071] was approved within the 3GPP forum in 1999 to
be mandatory for circuit- and packet-switched speech in UMTS networks. An AMR
speech codec adapts the error protection level to the local radio channel and traffic
conditions so that it always selects the optimum channel and codec mode to deliver
the best combination of speech quality and system capacity. AMR uses Multi-Rate
Algebraic Code Excited Linear Prediction (MR-ACELP) scheme based on two
different synthesis filters. It converts a narrowband speech signal (from 300 to 3,400
Hz) to 13-bit uniform Pulse Coded Modulated (PCM) samples with 8 kHz sample
rate. This leads to 20 ms AMR frames consisting of 160 encoded speech samples.
This means that the codec can switch mode, i.e. source bite rate, every 20 ms. AMR
has 8 coded modes in UMTS systems, whereas in GSM AMR uses either 6 or 8
modes. The eight source rates vary from 4.75 to 12.2 kbps. It also contains a low
rate encoding mode, called SIlence Descriptor (SID), which operates at 1.8 kbps to
produce background noise and a non-transmission mode.

The AMR codec dynamically adapts its error protection level to the channel
error conditions. For instance, lower speech coding bit rate and more error
protection schemes are used in bad channel conditions. This principle is illustrated
in Figure A1.1 where AMR strives to change to the best curve associated to a given
AMR mode. It has been shown that the degradation on the audio quality caused by a
lower speech coding rate is compensated by increased robustness with the channel
coding. Note, however, that this channel robustness is more beneficial in GSM than
in UMTS due to the embedded fast power control used in WCDMA systems. Using
a variable-rate transmission scheme also makes it possible to control the
transmission power of the UE, a fact that is particularly useful when the UE
376 UMTS

suddenly attains its maximum transmit power: in CDMA: lower bit rates generally
need lower transmit power and vice versa.

The “optimum” AMR codec


Speech quality mode is dynamically selected as
function of the channel quality
Good

Poor Channel quality


Good Bad

Figure A1.1. AMR principle

A1.1. AMR frame structure and operating modes

Figure A1.2 depicts the generic structure of the AMR frame. As observed in the
figure, the frame is divided into a header, auxiliary information and core frame. The
header contains the Frame Types and Frame Quality Indicator fields. The Frame
Type can indicate the use of one of the eight AMR codec modes for that frame, a
noise frame, or an empty frame. The Frame Quality Indicator indicates if the frame
is good or bad. The auxiliary information part includes the Mode Indication, Mode
Request and Codec CRC fields. The CRC field is used for the purpose of error-
detection calculated over all the Class A bits in the AMR Core frame. The Core
frame part is used to carry the encoded bits divided into A, B and C classes. In case
of a comfort noise frame, comfort noise parameters, i.e. a SID frame, replace “class
A” bits of the core frame while “class B” and “class C” bits are omitted.

AMR frame

Frame Type (4 bits)


Header
Frame Quality Indicator (1 bits)

Mode Indication (3 bits)


Auxiliary Information
Mode Request (3 bits) (for Mode Adaptation,
and Error Detection)
Codec CRC bits (8 bits)

Class A bits
Class B bits Core frame
(speech or comfort noise)
Class C bits

Figure A1.2. Generic structure of the AMR frame


AMR Codec in UMTS 377

Classification of the encoded bits according to their sensitivity to errors


AMR encoded bits are divided into three indicative classes according to their
importance: A, B and C. The reason for dividing the speech bits into classes is that
they can be subjected to different error protection in the network. Class A contains
the bits that are most sensitive to errors and any kind of errors in these bits typically
result in a corrupted speech frame which should not be decoded without applying
appropriate error concealment. This class is protected by the CRC in auxiliary
information field. Classes B and C contain bits where increasing error rates
gradually reduce the speech quality, but the decoding of an erroneous speech frame
is usually possible without a strongly perceptible quality degradation.

AMR operating modes


Table A1.1 depicts the 8 different modes (source bit rates) AMR can operate. It
should be noted that some of these modes are equivalent to the speech codecs
currently used in other mobile communication systems. For instance, the “AMR
12.20 kbps” mode is equal to the ETSI GSM called codec EFR (Enhanced Full Rate
Speech [TS 06.60]). Similarly, the “AMR 7.40 kbps” mode is equivalent to the
IS-641 codec used in the USA standard IS-136 (US TDMA). Finally, “AMR
6.70 kbps” mode is equivalent to the codec used in the PDC Japanese standard.

Frame type Frame content ClassA Class B Class C


index (AMR mode, comfort noise, or other) bits bits bits

0 AMR 4.75 kbps 42 53 0


1 AMR 5.15 kbps 49 54 0
2 AMR 5.90 kbps 55 63 0
3 AMR 6.70 kbps (PDC EFR) 58 76 0
4 AMR 7.40 kbps (IS-136 EFR). 61 87 0
5 AMR 7.95 kbps 75 84 0
6 AMR 10.2
. kbps 65 99 40
7 AMR 12.2 kbps (GSM EFR) 81 103 60
8 AMR SID – – –
9 GSM EFR SID – – –
10 TDMA EFR SID – – –
11 PDC EFR SID – – –
12-14 Future usage – – –
15 No data to transmit/receive – – –

Table A1.1. AMR modes and relationship with AMR frame structure
378 UMTS

Based on the fact that voice activity in a normal conversation is about 40%, all
AMR modes implement a Voice Activity Detection (VAD) algorithm that detects if
each 20 ms-frame contains speech or not on the transmitting side. VAD works
together with the Discontinuous Transmission (DTX) or Source Controlled Rate
(SCR) [TS 26.093] techniques where RF transmission is cut during speech pauses.
When the transmission is cut, “comfort noise” parameters are sent at a regular rate in
AMR frames during discontinuous activity. These frames are known as SID (SIlence
Descriptor) frames. The receiver decodes these parameters and generates locally a
“comfort noise”. Without this background noise the participants in a conversation,
might think that their connection is broken during silence periods. The SCR
technique for AMR in UMTS is mandatory and aims at prolonging the battery life
(UE side) and reducing the interference.

A1.2. Dynamic AMR mode adaptation

The AMR mode adaptation in UMTS networks means using different AMR
coding for the data stream. Mode adaptation can independently be applied in the
uplink and the downlink. At any point in time, a different AMR mode can be used in
each direction and this can be dynamically changed during a voice conversation.

Location of the AMR speech codec in UMTS networks


The AMR speech codec is located in the Transcoder (TC) function defined to be
in the UMTS core network and as such, logically controlled by Non-Access Stratum
protocols. From the transfer point of view, this means that all AMR coded data is
going to be transmitted not only via Iub and air interface but also via Iu-interfaces.
Note, however, that the AMR mode control that generates the AMR mode command
cannot be located in the TC, since this control entity needs information from the air
interface to make a decision about valid AMR modes – the AMR mode command is
used to change the current AMR mode to the new one. The only element in the
network which can provide this type of information is the UTRAN. Note that in
GSM networks the control of the codec mode is provided by the BTS. This solution
is not applicable in UTRA due to the soft-handover procedure defined for dedicated
traffic channels. Therefore, the AMR mode control function is part of the RNC, and
more precisely a part of layer 3 functionality. Within the radio interface, the rate on
the speech connection is either decreased or increased depending on the new valid
AMR mode by changing the valid Transport Format (TF) in the corresponding
MAC-d entity (see Chapter 7).

AMR mode adaptation in the downlink


As shown in Figure A1.3, the RNC generates the AMR mode adaptation
command based on existing radio conditions in the downlink as reported by the UE
AMR Codec in UMTS 379

from radio quality measurements and from traffic volume measurements. The
command is sent to the encoder inside the TC via the Iu interface.

Control
of AMR modes
UE Node B RNC TC
AMR speech Uu Iub Iu-CS AMR speech
codec codec

AMR encoded speech data in UL (ongoing call)


Request to modify Change
the AMR mode in DL AMR mode

AMR encoded speech data in DL with new AMR mode (ongoing call)
Change Request to modify
AMR mode the AMR mode in UL

AMR encoded speech data in UL with new AMR mode (ongoing call)

Figure A1.3. Overview of AMR codec mode control during an ongoing voice call

Uplink AMR mode adaptation


Two different alternatives for the AMR mode control in the uplink have been
proposed:
– Based on the air-interface load, the RNC decides when to request the encoder
in the UE to change the valid AMR mode and a new valid AMR mode is sent to the
UE inside the AMR mode command message. When received by the UE, mode
adaptation is made accordingly (see Figure A1.3). Within this approach, the UE
does not have any rights to request the mode adaptation from the network nor to
change the used AMR mode autonomously.
– In the second proposed alternative, the AMR codec control is not only included
into the RNC but also into the UE. This enables the UE to change the valid AMR
mode of the speech connection on uplink more quickly without requesting the mode
change from the RNC first. For instance, when the level of the maximum
transmission power is reached, the UE may change the valid AMR mode
independently. The new mode can, however, be selected only from the valid
Transport Format Set (TFS), which has been communicated to the UE by RRC from
the RNC side. The changed AMR mode is discovered by the RNC from the TFCI
bits in the uplink dedicated physical data channel.
380 UMTS

A1.3. Resource allocation for an AMR speech connection

An AMR speech connection can be initiated either by the UE or the network.


When the UE requests resources from the network, a first negotiation is made based
on NAS procedures in order to configure the call connection. The CN will determine
the QoS, needed which will be then indicated to the UTRAN inside the RANAP RAB
ASSIGNMENT REQUEST message. Based on this request, RNC can define the requested
RAB and associated Radio Bearer(s) (RB). Depending on whether the requested
AMR base speech connection supports the concept of Unequal Error Protection
(UEP) or Equal Error Protection (EEP), the RNC assigns either one or three RBs
(including one or three DCHs), respectively, for the user plane (see Figure A1.4). In
the control plane, RRC may allocate one or none signaling radio bearer according to
the alternative method used to change the AMR mode.

NAS radio bearer 1 radio bearer 2 radio bearer 3

81 bits 103 bits 60 bits


RLC
RLC (TM) RLC (TM) RLC (TM)

81 bits 103 bits 60 bits


MAC DTCH DTCH DTCH

81 bits 103 bits 60 bits

DCH (Class A bits) DCH (Class B bits) DCH (Class C bits)


TTI = 20 ms TTI = 20 ms TTI = 20 ms
Conv. coding 1/3 Conv. coding 1/3 Conv. coding 1/2
Physical CRC = 12 bits CRC = 0 bits CRC = 0 bits
layer
303 o 294 bits (RM) 333 o 324 bits (RM) 136 o 128 bits (RM)

DPCH Data 1 TPC TFCI Data 2 Pilots


6 2 0 28 4

DPCH slot structure (SF = 128)

Figure A1.4. Implementation of AMR 12.2 kbps mode in the radio interface

A1.4. AMR wideband

The AMR wideband (AMR-WB) codec has been standardized in 3GPP and is
part of the specification in Release 5. The codec is based on the same adaptive
AMR Codec in UMTS 381

principles as the AMR narrowband. The AMR-WB comprises nine codec modes:
6.6 kbps, 8.85 kbps, 12.65 kbps, 14.25 kbps, 15.85 kbps, 18.25 kbps, 19.85 kbps,
23.05 kbps and 23.85 kbps. The encoder of the AMR-WB is able to code an audio
signal with bandwidth between 50 and 7,000 Hz. A higher sampling rate is thus
needed compared with the narrowband approach (16 kHz instead of 8 kHz) leading
to a 14 bit samples with 16,000 samples/s. Wideband coding provides improved
voice quality especially in terms of increased voice naturalness since it covers twice
the audio bandwidth compared to the classical telephone voice bandwidth of 4 kHz
[TS 26.190, R5].

You might also like