Performance of Code Allocation Algorithms On UMTS Uplink With Mixed Voice/Data Traffic
Performance of Code Allocation Algorithms On UMTS Uplink With Mixed Voice/Data Traffic
Performance of Code Allocation Algorithms On UMTS Uplink With Mixed Voice/Data Traffic
ABSTRACT
Control Plane
In this paper, we are concerned with the performance evalua- User Plane
RRC
tion of code allocation algorithms in Universal Mobile Telecom- L3
nal, and the other based on dividing the available codes equally
among the requesting users. Further, for a data-only system, we Physical Layer L1
I. INTRODUCTION
nels [8]. For example, a 64 Kbps packet switched (PS) data trans-
UMTS networks will offer multiple services (voice, data, port channel and a 12.2 Kbps circuit switched (CS) voice trans-
video, etc.) with different quality-of-service (QoS) requirements port channel can be multiplexed on to a 128 Kbps physical chan-
to mobile users [1]-[3]. UMTS networks support high speed ra- nel. Transport channels are defined by how and with what fea-
dio access (up to 2 Mbps) based on Wideband Code Division tures (e.g., transport block size, transmission time interval, etc.,
Multiple Access (WCDMA). The users, during call establish- which are defined in the following paragraph) data is transferred
ment, request the network the desired QoS for the connection over the air interface.
in terms of data rate, delay, priority, reliability, etc. The network
allocates the available resources (transmission rate and transmit Refer Fig. 2 which illustrates the following definitions. The
power) to different users based on certain resource allocation pol- basic unit of data exchange between L1 and MAC for L1 pro-
icy. Rate allocation to each user can be varied either by allocat- cessing is called a transport block (TB). The transport block size
ing multiple spreading codes with constant spreading factor or by is the number of bits in a TB. A set of TBs exchanged between
varying the spreading factor. In WCDMA, the physical channels L1 and MAC during the same frame using the same transport
follow a layered structure of radio frames and time slots. Each channel is called a transport block set (TBS). The TBS size is the
radio frame is of 10ms duration and consists of 15 slots. The number of bits in the TBS. The TB size and the TBS size are cho-
maximum number of bits sent per radio frame depends on the sen by the MAC at the user terminal. The periodicity at which a
rate allocated to the user. The network can dynamically vary the TBS is transferred by L1 on to the air interface is called transmis-
rate allocation to different users on a frame-by-frame basis. sion time interval (TTI). The allowed values of TTI are 10, 20 40,
The radio interface protocol layers which are involved in the and 80 ms. The MAC delivers one TBS to L1 every TTI. The for-
radio resource allocation are shown in Fig. 1 [4]. The radio re- mat (TB size, TBS size, type/rate of coding, size of CRC) offered
source control (RRC) in Layer 3 (L3) is responsible for the sig- by L1 to MAC (and vice versa) for the delivery of TBS during a
naling and control information exchange between the user and TTI on a given transport channel is called transport format (TF).
network to effect allocation and deallocation of radio resources The transport format determines the transport channel bit rate be-
[5]. The data transport services offered by the physical layer fore L1 processing. For example, let TB size = 336 bits (320 bits
(Layer 1 – L1) [6] is achieved through the use of transport chan- payload + 16 bits header), TBS size = 2 TBs per TTI, and TTI =
nels via the media access control (MAC) [7] sublayer in Layer 10 ms. Then the transport channel bit rate (with header) is given
2 (L2). Each user terminal may simultaneously have multiple by 336*2/10 = 67.2 Kbps. The user bit rate (without header) is
transport channels multiplexed on to one or more physical chan- given by 320*2/10 = 64 Kbps. Thus, it can be seen that vari-
able bit rate transmission on a single transport channel can be
This work was supported in part by L & T Infotech Ltd., Bangalore 560001 achieved by changing (from one TTI to the other) either the TBS
under project No. SID/PC14032. size only, or both the TB size and the TBS size. A set of transport
formats associated with a transport channel is called a transport A. Code Allocation Algorithms
format set (TFS). A combination of TFs on different transport The network allocates channel rates to users by assigning dif-
channels in a given TTI is called transport format combination ferent number of spreading codes with different spreading fac-
(TFC). A set of transport format combinations allowed by the tors. The spreading factor can be in the range 4 to 256 on the
network is called the transport format combination set (TFCS). UMTS uplink. In other words, with a 3.84 Mcps chip rate, the
The network informs the TFCS to the user terminal to be used minimum channel rate can be "!#%$&')(* & Kbps. This
on the uplink transmission by the user terminal. The MAC at channel rate will typically be larger than the user information rate
the user terminal then chooses between the different TFCs spec- because of overhead bits due to coding, CRC, etc.
ified in the TFCS. The user terminal MAC can (based on certain
criteria) choose different TFCs for different TTIs. In our model, we consider that the network allocates channel
rates to users in integer multiples of +), , where +-, is the
Transport Format Combination (TFC)
minimum channel rate that can be assigned to a user. Here, we
take + , .(/' Kbps which corresponds to a spreading fac-
tor of 64. This 60 Kbps channel rate can carry a 12.2 Kbps CS
TB voice call or 16 Kbps PS data traffic, both with their associated
Transport TB TB TB
Channel 1 overhead bits due to coding, CRC, etc [9]. The values of the
TTI
available channel rates then are 60, 120, 180, and 240 Kbps. We
Transport Format Set 10
(TFS) assume that spreading codes each corresponding to rate +2,
Transport Block are available with the network for allocation. The network can
Set (TBS)
Transport Block change the allocation of spreading codes to different users ev-
TB
Transport TB TB TB ery TTI. We assume that the network gives priority to voice calls
Channel 2
while allocating spreading codes, i.e., available codes are first al-
Transmission Time Interval
(TTI) Transport Format lotted to voice calls and the remaining codes are allotted to data
(TF)
Transport Format Combination Set (TFCS)
calls. The algorithms presented in the following are for assigning
codes for the data calls. A code allotted to a voice call is held
for the entire call duration. In user terminals with simultaneous
Fig. 2. Transport channel definitions voice and data calls, silence periods in voice can be used to send
data packets. For terminals with only data calls, codes can be
In this paper, we are interested in evaluating the performance allocated/deallocated every TTI.
of different code (rate) allocation algorithms on the UMTS up- A.1 Algorithm I
link in a mixed voice/data traffic scenario. Two different code
allocation algorithms are considered: one based on the overall In this algorithm, the network allots codes to users based on
buffer occupancy at the user terminal, and the other based on di- the data buffer occupancy at the individual user terminal. The
viding the available codes equally among the requesting users. In network allocates codes in such a way that the user with a larger
the buffer occupancy based algorithm, the rate allocation is done buffer occupancy will be assigned a higher rate.
is such a way that the user with a larger buffer occupancy will be 30
Let be the number of users in the system. Let be the total
assigned a higher rate. We also evaluate the performance of SIR
number of codes of rate corresponding to +4, available in the
based rate and power allocation algorithms in a data-only system. 90
system. Let , 65785 , be the maximum number of codes
of rate corresponding to +9, that can be allotted to a given user.
II. SYSTEM MODEL
It is assumed that the network has the knowledge of buffer occu-
We consider a mixed voice/data system in which there are pancy in all the participating user terminals. The network first
mobile user terminals. Each user can generate a CS voice call assigns codes to newly arrived voice calls. The network then ar-
and/or a PS data call. That is, at any given time, a terminal can ranges the users in descending order of data buffer occupancy.
have a) neither a voice nor a data call, b) a voice call and no data Let be the number of transport blocks in the buffer of the :%;
call, c) a data call and no voice call, and d) simultaneously both user in
the ordered list. Let < be the number of codes allotted to
a voice as well as a data call. Voice call arrivals are assumed the :=; user in this list. Note that < can be either 1 or 0 depend-
to follow a Poisson distribution, i.e., the voice call inter-arrival ing on whether the :; user has a voice call or not. The number
time is exponentially distributed with mean . The voice call of the remaining codes available for allocation to data calls (in-
holding time isalso
assumed to be exponential with mean call cluding data calls in user terminals with simultaneous voice and
holding time . Since speech consists of alternating active and data), >@? , is then given by
silence periods, we model the voice source as a ON/OFF source 10BADE C
where the ON and OFF periods are assumed to be exponentially >
?6( < (1)
F@G
distributed with means
and
, respectively. During a PS
data call, packets arrive in bursts where the burst inter-arrival
time The algorithm performs code allocation to the largest buffer
is assumed to be exponentially distributed with mean , and occupancy user first, the second largest buffer occupancy user
the number of transport blocks (TB) per burst is assumed to be next, and so on. This process is continued until the least buffer
geometrically distributed with mean . occupancy user gets the code allocation or the available codes are
exhausted, whichever happens first. Accordingly, the number of TB size, bits 336
(( ''
TTI, ms 20
codes assigned to the : ; user in the list, , is obtained as
(2)
TF0 0 336
(( ''
A TF1 1 336
G
( : > < :
( $
where
>
:B( $
codes have been allotted to the :; user in the list. With the
TF3
TF4
3 336
4 336
TABLE II
( '
definition of >B? in Eqn. (1) as the initial condition, the > s are T RANSPORT CHANNEL PARAMETERS FOR 64 K BPS
recursively obtained using the relation
INTERACTIVE / BACKGROUND DATA TRAFFIC
>
( >
G
A
(3) for the mixed voice/data traffic scenario model that we have con-
sidered. It is seen that voice call is allotted one code and data call
A.2 Algorithm II
can be allotted up to four codes. It is noted that the combination
In this algorithm, the network allots codes to different users TFC10 allows more bits to be sent per code allotted through the
based on dividing the available codes equally among the request- use of increased puncturing (puncturing limit = 0.76) [9].
)+*,.-
ing users. Like in Algorithm I, codes are allocated to voice calls
first. Following the same notation used in Algorithm I, the as- Voice Trch2 Data Trch No. of
signment of the remaining codes to the data calls is done as fol-
lows.
TFC1 TF0 && , TF0 && , TF0 TF0 && (( codes required
0
TFC2 TF0
&& , TF0
&& , TF0 TF1
&& (( 1
TFC3 TF0 , TF0 , TF0 TF2 2
buffer occupancies,
TFC6
TFC7
TF1
TF1 && , TF1
, TF1 && , TF1
, TF1
TF0
TF1 && (( 1
2
( :
!!"
A
+
:
G
A
A
<
$#%
:@(
$ T RANSPORT FORMAT COMBINATION SET
TABLE III
(TFCS) FOR MIXED VOICE / DATA
TRAFFIC
0/
(4)
16
-
T_i=120s. Algo I meet a given BER, the received # should be greater than a
T_i=120s. Algo II
Average data burst delay (frames) 250 T_i=180s. Algo I given threshold value, .
;
T_i=180s. Algo II
T_i=240s. Algo I
T_i=240s. Algo II A. SIR based algorithm with power control
200 T_i=360s. Algo I
T_i=360s. Algo II The user terminals with backlogged data packets are ordered
T_i=600s. Algo I
150 T_i=600s. Algo II according to their received SIR at the base station. The code allo-
T_i=1500s. Algo I cation is then done in descending order of the received SIR, i.e.,
T_i=1500s. Algo II
the codes are allotted to the user with the highest SIR first, and so
100
on. The SIR estimates of user terminals which transmitted pack-
ets in the previous frame can be measured on the traffic channel
50
and made available at the base station. The SIR estimates of the
user terminals with new message arrivals can be measured from
0 the resource request transmission made by them on the random
0 0.5 1 1.5 2 2.5 3 3.5 4
access channel. The resource request transmission for new mes-
/ /
Fig. 3. Average data burst delay vs. data burst arrival rate. . . power, > , .
sage arrivals is done using the maximum permissible transmit
as the voice call arrival rate increases (e.g., ( $ secs), the
2/10
,
> > :
; -
F
IV. SIR BASED ALGORITHMS
In the previous section, the code allocation algorithms did not > 5 > , (10)
7- "
at the base station for a data-only system. In the first algorithm, E
+
,
the user terminal is allocated a transmission rate as well as trans- > (
-
> : (11)
; F
mit power by the base station. In the second algorithm, the base
station allocates only the transmission rate and the user terminal In matrix notation, 9+:<; %=:
transmits at the maximum allowable transmit power, > , . In ( (12)
J
A
-
A
-
user. The received SIR at the base station for mobile : is given by
C AC J C D E U """ G
@ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .HG . . . . . . . L
A G A
I
-
M C """ C A J G D E U
>
(5) + ( A
A
G
C
>
for mobile , and is the thermal noise power. The channel gain,
> :
nals which have been a code in this iteration.
, is given by ! " %$ J'&
:
#
-
G
?
If the power allocation to any user terminal exceeds the max-
-
(6)
( imum permissible transmit power level, > , , its rate of trans-
where is the distance of mobile from the base station, is
mission is reduced and the power vector is again calculated. This
the path loss exponent, and ( is the normally distributed shadow
:
process is continued till a feasible power vector is found.
fading random variable for mobile . The received )+* for : #
B. SIR based algorithm without power control
) *
,
mobile is then given by, :
In this case, the user terminals are allowed to transmit with the
(7) ( +
maximum power level, > , , and no power optimization is done
+
at the base station. The requesting user terminals are arranged It is observed that the algorithm with power control ensures that
in descending order of received SIR. Maximum possible rate is nodes which are far away see the same delay as the nodes which
allotted to the user terminal with the best received SIR. Rate is
then assigned to the user terminal with the next best SIR, and nodes transmit at > ,
2
are closer to the base station. In the no power control case, all the
resulting in a high interference from the
so on till either the codes are exhausted or all the user terminals nearby nodes. This causes the far off nodes to wait till the end
)*
have been assigned codes. The base station then calculates the of the transmission from closer nodes before being able to send
received # for each user terminal using this rate allocation
2 )<*
their data bursts. Hence, the far off nodes experience a higher
-
and transmit power > , . If the received # for any user delay than the nearby nodes.
terminal is less than , the rate allotted to that user terminal is
;
reduced till the received # exceeds . ) * - ;
50
With power control
V
Without power control
/
20 dBm
(
12 35
6 16
9 dB
30
TABLE V 25
S IMULATION PARAMETERS FOR SIR BASED ALGORITHMS
20
Distance (m)
and variance
and is correlated with correlation of . The
Fig. 5. Average data burst delay vs. distance. . . Data burst
simulation parameters are given in Table V. The path loss model arrival rate = 1 message/sec/node.
Q
" Q
is taken to be
V. CONCLUSION
( $
'
G
? + (13)
where R is the distance of the user terminal from the base station In this paper, we evaluated the performance on two code al-
in km [10]. location algorithms for mixed voice/data traffic on the UMTS
uplink. Two different code allocation algorithms were consid-
40 ered: one based on buffer occupancy at the user terminal, and the
With power control other based on dividing the available codes equally among the
Without power control
Average data burst delay (frames)
REFERENCES
(
15
[1] H. Holma and A. Toskala, WCDMA for UMTS: Radio Access for Gen-
10 eration Mobile Communications, John Wiley & Sons, 2000.
0 0.5 1 1.5 2 2.5 3 3.5 [2] H. Kaaranen et al, UMTS Networks, Architecture, Mobility and Services,
John Wiley & Sons, Chichester, 2002.
/ /