. 67
( 132 .)


Figure 6.57 shows an example of a segment of data passing through the MAC and
RLC protocol stack. In this example, the RLC unacknowledged mode is used, and the
DTCH/DCCH is mapped to the FACH. At the RLC layer, the data is segmented and a
sequence number and length ¬eld are added to it. At the MAC layer, the UE is identi¬ed
on the FACH using its U-RNTI or C-RNTI (see Section 6.16.2). The TCTF will have the
value 11, which indicates that this is a DTCH (or DCCH) rather than a BCCH, CCCH
or CTCH.

6.12.5 MAC and physical layer interaction
The following are de¬nitions applied to data that is transferred between the MAC and
physical layers. They describe the format and rate of the data.

• Transport block (TB): this is the basic unit exchanged between the physical layer and
the MAC. At the physical layer, a CRC may be appended to each TB.

• Transport Block Set (TBS): a group of TBs, passed between the physical layer and the
MAC during a given time interval and using the same transport channel.
• Transport block size: the number of bits in a given TB. All TBs in a TBS must have
the same size.
• Transport block set size: the number of bits in a TBS. It is equal to transport block set
— transport block size.
• Transmission time interval (TTI): the time period during which a TBS is transferred
across the radio interface. The MAC layer will deliver one TBS, i.e. one or more TBs,
to the physical layer every TTI. The TTI is always a multiple number of radio frames
(one radio frame is 10 ms.) An example of the MAC/physical layer exchange for two
dedicated channels is shown in Figure 6.58. It can be seen that the data rate can be
varied in every TTI by changing either the size of each TB or the number of TBs that
are sent.

Transport format
The transport format describes the structure of the TBS that the physical layer will deliver
in a TTI. It is broken into two parts, the dynamic part and the semi-static part:

• Dynamic part: this consists of the transport block size and the transport block set size.
• Semi-static part: this consists of the TTI, the error protection scheme (turbo, convo-
lution or none, the coding rate, and the rate matching parameter), and the size of the
CRC. This is set up via the upper layer RRC signalling and thus to modify requires
further signalling between the UE and the RNC.

Consider that there are four TBs of size 336 bits, then the transport format could be
de¬ned as:
Dynamic part: 336 bits, 1344 bits
Semi-static part: TTI = 20 ms, turbo coding, 1/3 rate, rate match = 1, CRC = 16 bits






Figure 6.58 Example of transport blocks

Transport format set
The transport format set (TFS) describes all transport formats that are possible on a
channel. This is to support variable data rates, since the rate can change every TTI. For a
TFS, the semi-static part will be the same for all formats. For example, a TFS could be
described as:
Dynamic part: (336 bits, 336 bits), (336 bits, 672 bits), (336 bits, 1008 bits),
(336 bits, 1344 bits)
TTI = 20 ms, turbo coding, 1/3 rate, rate match = 1,
Semi-static part:
CRC = 16 bits
Here, the TB size remains the same for each TTI, but the number of blocks can vary
between one and four. Consider that this uses RLC acknowledged mode with no MAC
layer multiplexing, this accounts for 16 bits of overhead, meaning the data is actually
320 bits. This then corresponds to data rates of 16 kbps (320/20 ms), 32 kbps, 48 kbps
and 64 kbps, and to transfer the data would take 80 ms (4 — TTI), 60 ms, 40 ms and
20 ms, respectively. This allows the data rate to vary every TTI without any need for any
signalling between the UE and the network.

Transport format combination
The physical layer may take one single transport channel and place it into a physical
channel, or it may multiplex several transport channels together. Each transport channel
will have a TFS associated with it. However, if several are being multiplexed together,
at a given instance, not all may be present. The transport format combination (TFC) is
an authorized combination of transport formats that may be passed to the physical layer.
This means that not all permutations from the TFS of the transport channels may be
allowed. This could be used, for example, to prevent all multiplexed channels using their
maximum rate simultaneously. The resulting channel consisting of several multiplexed
transport channels is referred to as a coded composite transport channel (CCTrCH).

Transport format combination set
A transport format combination set (TFCS) is de¬ned as a set of TFCs on a CCTrCH.
It is the MAC layer that will decide which TFC from the TFCS to pass to the physical
layer. However, the assignment of the TFCS is performed during bearer establishment
signalling procedures, as described in Section 6.16.7, prior to transfer of data.

Transport format indicator
The transport format indicator (TFI) is a label for a particular transport format in a
transport format set. Every time a transport block is passed between the MAC and the
physical layer, a TFI for that transport block is also exchanged.

Transport format combination indicator
The physical layer may receive several transmission blocks that it will multiplex into a
CCTrCH. Each block submitted by the MAC will also be associated with a TFI. The

TrCH 1 TrCH 2

Physical Layer


Figure 6.59 Transport format combination

physical layer will then generate a transport format combination indicator (TFCI) to
indicate this current transport format combination. The TFCI is included in the physical
control channel associated with the data to inform the receiver of the transport format in
the current transmission.
Figure 6.59 shows an example of two transport channels (TrCH) being passed to the
physical layer for transmission on a single dedicated data channel (DPDCH).
The TFI/TFCI part is optional and does not need to be transported if blind transport
format detection (BTFD) is employed, where the receiver can work out the transport
format using other information such as CRC, DPDCH/DPCCH power ratio, etc.
As an example of how all the de¬nitions ¬t together, consider that a user is making
a phone call using AMR 12.2 kbps, while also browsing the web. For the web traf¬c,
a maximum rate of 64 kbps has been allocated, with intermediate rates of 16 kbps and
32 kbps. The TFS for the channels could look as follows:
TFS for TrCH 1 (12.2 kbps AMR)
Dynamic part: 244 bits, 244 bits
TTI = 20 ms, convolution coding, 1/3 rate, rate match = 1,
Semi-static part:
CRC = 12 bits
TFS for TrCH 2 (64 kbps data, with 16 kbps and 32 kbps rates)
Dynamic part: (320 bits, 320 bits), (320 bits 640 bits), (320 bits, 1280 bits)
TTI = 20 ms, convolution coding, 1/3 rate, rate match = 1,
Semi-static part:
CRC = 16 bits
Dynamic part:
Combination 1 “ TrCH 1: (244, 244) TrCH 2: (320, 320)
Combination 2 “ TrCH 1: (244, 244) TrCH 2: (320, 640)
Combination 3 “ TrCH 1: (244, 244) TrCH 2: (320, 1280)
Semi-static part:
TrCH 1: TTI = 20 ms, convolution coding, 1/3 rate, rate match = 1, CRC = 12 bits
TrCH 2: TTI = 20 ms, convolution coding, 1/3 rate, rate match = 1, CRC = 16 bits
Figure 6.60 illustrates the rates and possible combinations for these two channels. A
value of 0 kbps indicates that there is no data being sent on the channel (discontinuous
transmission; DTX).

TrCH 1 TrCH 2
Combination TFCI 6

12.2kbps 16kbps

0kbps 0kbps
Format Set

Figure 6.60 TFCS example

In practice, to provide a given user data rate requires the transport block size to include
the RLC overhead, which is dependent on the chosen RLC mode of operation. Consider
that the user application requires 64 kbps, and the RLC unacknowledged mode. If the TTI
is 20 ms, then the application will send 320 bits each TTI. However, the RLC will add
2 bytes of overhead (7-bit sequence number + 7-bit length ¬eld + 2 — E bits). Therefore,
the TB size will be 336 bits. Since 12.2 kbps speech is transferred in RLC transparent
mode there is no additional overhead added for this channel.


The previous example used the basic rate of 12.2 kbps for a voice call. It showed that this
equated to a 244-bit transport block which is sent each TTI = 20 ms. However, due to the
operation of the AMR CODEC in UMTS, the situation is somewhat more complex, but
it serves as a useful example of the interoperation of the radio link protocols. The AMR
speech CODEC is a multirate speech coding scheme consisting of eight source rates from
4.75 kbps to 12.2 kbps. Table 6.11 shows the rates.
These rates are CODECS from various sources; for example, the AMR 12.20 is the
GSM Enhanced Full Rate (EFR) and the AMR 6.70 is the EFR CODEC as used in
the PDC network in Japan. The AMR SID is a silence descriptor rate for encoding of
low-rate background noise and consists of 39 bits sent every TTI.
The rationale for support of a number of rates is that for traf¬c management, the source
rate being used can be changed in the downlink when traf¬c on the air interface exceeds
an acceptable load or when the link is exhibiting poor quality. In the uplink, it can also
be changed because of loading considerations or to extend uplink coverage. In theory, the

Table 6.11 AMR CODEC rates
CODEC mode Bit rate
AMR 12.20 12.20
AMR 10.20 10.20
AMR 7.95 7.95
AMR 7.40 7.40
AMR 6.70 6.70
AMR 5.90 5.90
AMR 5.15 5.15
AMR 4.75 4.75
AMR SID 1.80

rate can be changed every 20 ms (the TTI). However, in practice the adaptation would
need to occur much less often.
The CODEC divides bits up into three different classes, classes A, B and C, according
to their relative importance within the coding scheme. Class A bits are the most important
and have a high level of error protection, including a CRC check. These bits, if corrupted,
will result in audible degradation of speech quality. Erroneous speech frames are thus
identi¬ed by the CRC check, and should have some error concealment techniques applied
to them. For class B and C bits, their importance is less, with class C being less sensitive
to errors than class B. These are afforded less error protection. Class B and C speech
frames containing errors can be passed on. The voice quality will degrade with increased
error content; however, the effects of errors are much less signi¬cant than for class A.
The number of bits per class for each AMR rate is shown in Table 6.12. Note that not
all rates have class C bits.
The CODECs are all based on the algebraic code excited linear prediction (ACELP)
algorithm, which assumes that the vocal tract is a linear ¬lter receiving inputs in the form
of air from the lungs being vibrated at the vocal cords to produce speech. This means that
the audio signal can be compressed by a high factor; however, the compression algorithm
is optimized for voice.
It was seen in Chapter 2 that although in standard telephone networks, voice is allocated
resources as a full-duplex application, in fact voice activity is closer to half-duplex, as

Table 6.12 AMR CODEC bit numbers
CODEC Class A Class B Class C Total
AMR 12.20 81 103 60 244
AMR 10.20 65 99 40 204
AMR 7.95 75 84 0 159
AMR 7.40 61 87 0 148
AMR 6.70 58 76 0 134
AMR 5.90 55 63 0 118


. 67
( 132 .)