Professional Documents
Culture Documents
Lte Tput WP 08july2016
Lte Tput WP 08july2016
Jari Salo
A frequently asked question among LTE radio planners is that of how to determine the maximum acceptable
LTE radio interface load that should not be exceeded in order to maintain some targeted user data throughput.
This white paper summarizes some simple formulas for calculation of downlink user throughput as a function
Physical Resource Block (PRB) utilization. The formulas are expressed in terms of standardized 3GPP KPIs and are
hence computable from network performance counters. Examples from live LTE networks are given to illustrate
the usefulness.
1
2 J. Salo
source block (PRB) utilization by means of the sence of other users, a single user obtains
well-known M/G/1 processor sharing formula, all available radio resources. For example,
resulting in 11 degradation in user through- constant bit rate streaming traffic would
put with denoting PRB utilization. Live net- not satisfy this condition.
work examples are shown to validate the result.
If there is more than one user, the sched-
The basic M/G/1 PS formula can be extended
uler shares radio resources equally, on av-
to cover the case where voice and data traffic
erage, between users. This fair sharing
are mixed. Other interesting use cases include
principle is assumed independently of the
external rate limitation and carrier aggregation
radio conditions of the UEs sharing the
which are discussed in Section 4. Further food
scheduler resources.
for thought is served in Section 5 where the va-
lidity of the underlying assumptions are scruti- The first assumption will be relaxed in Sec-
nized. Finally, conclusions are presented. tion 4, and the second assumption will be dis-
cussed in Section 5. The notion of "radio re-
Version: 8 July 2016
source" could be defined in various ways, but in
case of the LTE radio interface it is convenient to
Abbreviations
choose Physical Resource Blocks (PRBs) as the
FDD Frequency Division Multiplexing resource being shared. For LTE downlink, PRB
IP Internet Protocol utilization can be equated with transmit power
KPI Key Performance Indicator utilization as long as physical layer and com-
LTE Long Term Evolution mon channel overhead are properly taken into
M/G/1 Memoryless/Generic/single server account.
M/G/R Memoryless/Generic/R servers With the assumptions above, instantaneous
OSS Operations Support System user throughput with x active users download-
PRB Physical Resource Block ing simultaneously would be 1x of the maximum
PS Processor Sharing throughput. A UE is said to be active if there
RRC Radio Resource Control are data remaining in the transmit buffer1 . As
TCP Transmission Control Protocol different UEs in the cell start and finish their
TDD Time Division Multiplexing data transfers, the number of active UEs (x), and
TTI Transmission Time Interval hence also the instantaneous user throughput,
UE User Equipment changes over time. The interesting metric is the
average throughput experienced by UEs in the
2. THROUGHPUT VERSUS NUMBER OF cell.
ACTIVE USERS 2.2. Two user throughput metrics
In this section two different single-user Consider a UE located somewhere in a cell
throughput metrics are defined. It is shown via experiencing certain radio quality. In the ab-
a live network example that the so-called sched- sence of any other users the UE is allocated all
uled IP throughput, standardized by 3GPP, can available PRBs and receives some throughput
be accurately predicted based on the number of T1 , where the subscript 1 emphasizes that there
active UEs. is one active user, i.e., x = 1. If there were x > 1
active UEs in the cell, the throughput of the user
2.1. Basic Assumptions 1 TermsUE and user are used interchangeably. The num-
The following assumptions are made. ber of active UEs is different from the number of RRC-
connected UEs since a UE can be RRC-connected without
Full buffer traffic model so that, in the ab- having any data to receive or send.
LTE Radio Load versus User Throughput (DRAFT) 3
would be T1 /x instead. The maximum achiev- IP network throughput analysis where it goes
able user throughput, T1 , depends on the user by the name "flow throughput". An application
location in the cell, interference from other cells, to wireless network setting can be found [2] and
number of transmit antennas, and so on. The many others published since.
average user throughput Tue is defined as the ex- Regardless of which variant is used it is worth
pected value of T1 /x for positive integer x, in noting that the scheduled throughput Tsch is al-
other words ways lower than user throughput Tue . This is a
[ ] direct result of the concavity of 1/x and Jensens
T
Tue = E 1 , x 1 , (1) inequality: E[1/x ] > 1/E[ x ] and subsequently
x
Tue >Tsch . Typically one would be more inter-
where E[] denotes expected value and T1 and ested in the end user experience making Tue
x 1 are the random variables being averaged. preferable to Tsch . Despite of this the focus in
The user radio conditions, and hence the max- this paper is on the scheduled throughput Tsch
imum throughput T1 , can be assumed statisti- because Tue is not usually computable from LTE
cally independent of the number of active UEs base station performance counters. A discussion
in the cell, and (1) can thus be written as on the differences between different throughput
[ ] metrics can be found in [6,8].
1 Fig. 1 shows an example of two cells serv-
Tue = E[ T1 ] E , x 1. (2)
x ing a large number of smart-phone users. The
horizontal axis is the average number of active
The term C = E[ T1 ] will be called cell capacity
UEs, E[ x ], that has been extracted from hourly
in this paper.
OSS performance counters over a period of two
Unfortunately, the second term E[1/x ], which
weeks. The hourly averages of x have been
is the average of the inverse of the number of
binned to integers and for each bin the average
active UEs, cannot be always computed since it
flow throughput is plotted. The flow through-
is not usually available as a radio counter. On
put shown on the vertical axis is also obtained
the other hand, the average active UEs, E[ x ], is
from the OSS counters and computed according
a standardized KPI defined in 3GPP TS 36.314
to the scheduled throughput (Tsch ) definition in
and thus commonly implemented in commer-
3GPP TS 36.314. Each dot in the figure is the av-
cial systems. Therefore a more practical metric
erage UE throughput for the horizontal binned
results if E[1/x ] in (2) is replaced by 1/E[ x ], or
value of E[ x ]. It can be seen that the scheduled
E[ T1 ] throughput scales approximately inversely pro-
Tsch = , x 1. (3) portional to the average active UEs, as predicted
E[ x ]
by theory.
In 3GPP TS 36.314 this is called "scheduled From here onwards, unless otherwise men-
IP throughput". It should be emphasized that tioned, the terms scheduled throughput, flow
E[ x ] = E[1/x ] and for this reason Tsch and Tue throughput, user throughput, and UE through-
are different throughput metrics and not equal put are all used interchangeably to refer to Tsch .
in value. The scheduled throughput can also be
written as [6] 3. THROUGHPUT VERSUS PRB UTILIZA-
S TION
Tsch = , (4)
W In the previous section, it was shown that av-
where S is the average file size (bytes) and W is erage scheduled throughput is inversely propor-
the average file transfer time. This form of the tional to the average number of active UEs. For
scheduled throughput is often used in fixed-line most radio planners the number of active UEs
4 J. Salo
Scheduled throughput versus number of active UEs utilization is available from system counters in
100
any practical LTE system.
Scheduled throughput, Megabits per second
Comparison of user throughput and scheduled throughput utilization is extracted from hourly counter
1
user throughput Tue measurements collected over a period of two
Throughput as a fraction idle cell throughput C
15 30
10 20
5 10
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Megabits per second
30 20
15
20
10
10
5
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Megabits per second
10 10
5 5
0 0
0 20 40 60 80 100 0 20 40 60 80 100
PRB utilization, (percents) PRB utilization, (percents)
Figure 3. Scheduled throughput versus PRB utilization, six cells from three different networks.
LTE Radio Load versus User Throughput (DRAFT) 7
Scheduled throughput using M/G/R PS for different R Scheduled throughput versus PRB utilization, M/G/R PS
1 50
0.8 40
0.7 35
R = 1, 1.2, 1.5, 2, 3, 5
0.6 30
0.5 25
0.4 20
0.3 15
R = 1.7
0.2 10
0.1 5
0 0
0 20 40 60 80 100 0 20 40 60 80 100
PRB utilization, (percents) PRB utilization, (percents)
Figure 4. Normalized scheduled throughput Figure 5. Scheduled throughput versus PRB uti-
versus PRB utilization, M/G/R PS model. . lization, M/G/R PS model. .
mula (R = 1) as well as the M/G/R based for- traffic. The approximation arises from the fact
mula (9) for R = 1.7 and C = 47 Mbps. It can be that the number of PRBs allocated for the GBR
seen that the M/G/R PS formula offers better traffic is time-varying (but approximated with
match to network-measured throughput values. its average value) since it depends on the num-
ber of GBR users and their radio quality. The
4.2. Priority-based scheduling of non-GBR quasi-static approximation is still usable if the
data and GBR traffic average GBR PRB utilization is less than, say,
A mix of Guaranteed Bit Rate (GBR) traf- 50% and varies slower than non-GBR utilization
fic and non-GBR elastic data traffic is a practi- [12]. This is a viable assumption since voice call
cal use case due to voice and video over LTE. holding time is usually in the order of 60 sec-
The main question is how the non-GBR data onds or more while non-GBR data transfers of
throughput changes due to GBR traffic that is smart phones last considerably shorter time.
scheduled with strict priority over the data (pos- As a summary, when GBR traffic is scheduled
sibly up to some capacity limit). Although a with strict priority over non-GBR traffic an ap-
considerable amount of research has been con- proximate way to calculate scheduled through-
ducted, no exact formulas are available for the put is to replace the PRB utilization in (8) with
case of two or more traffic classes scheduled the effective non-GBR PRB utilization eff , which
with strict priorities. Bounds and approxima- can be calculated as utilized non-GBR PRBs di-
tions can be found in e.g. [12,13] and refer- vided by the number of non-GBR PRBs available
ences therein. For daily radio planning work, after GBR scheduling.
the only usable approach seems to be the so-
called quasi-stationary approximation that as- 4.3. Carrier aggregation
sumes that the capacity available to non-GBR For carrier aggregation, the total user
traffic is simply the difference of the total num- throughput is the sum of component carrier
ber of PRBs and average PRBs allocated for GBR throughputs. For the simplest case of two car-
LTE Radio Load versus User Throughput (DRAFT) 9
riers, this leads to leaves the portion of 1 for the second cell.
For a given average file size S, the average sector
user throughput (8) can be written as a weighted
Tsch,ca = Tsch,1 + Tsch,2 sum
= C1 (1 1 ) + C2 (1 2 ) , (13)
where Ci and i are the average link throughput Tsec = Tsch,1 + (1 ) Tsch,2 (15)
and load of the ith link, respectively. This is eas- S S
ily generalized to multiple carriers. From (13) = + (1 ) (16)
W1 W2
it can be seen that the contribution of a highly = C1 (1 1 ) + (1 )C2 (1 2 ) , (17)
loaded carrier on total throughput is very small.
The cell throughputs Ci are not, strictly speak- where Ci and i are the average cell throughput
ing, independent since due to higher (lower) and load of the ith cell, respectively. The traffic
path loss a UE on a higher (lower) carrier fre- splitting ratio is assumed to apply to all traffic
quency will in general be closer to cell edge. and thus the cell loads are3
Thus in a given UE location C2 depends on
C1 experienced in the same location. Assum-
S
ing that the dependency is linear, C2 = kC1 , 1 = , (18)
for some constant k, the carrier aggregation C1
throughput gain from adding the second carrier S
2 = (1 ) . (19)
would be C2
traffic split versus throughput, C1=20Mbps, S = 10Mbps Optimum traffic balancing factor for two cells
30 0.5
sector average user throughput, Mbps
25
0.4
0.2
10
0 0
0 0.2 0.4 0.6 0.8 1 1 1.5 2 2.5 3
traffic balancing ratio, ratio of cell capacities C /C
2 1
Figure 6. Average sector user throughput as a Figure 8. Optimum traffic splitting factor for
function of traffic split. C1 = 20Mbps, sector two parallel cells.
offered traffic is 10Mbps.
0.9
proportion of traffic carried by the first cell,
15
0.8
15
10
0.7 20
0.6
0.5
25
15
20
0.4
10
5
0.3
0.2
25
0.1
15
20
10
0
0
0 5 10 15 20 25 30 35 40
capacity of the second cell C , Megabits per second
2
Figure 7. Equal throughput contours for two cells for different traffic balancing ratios. The first cell
capacity is constant C1 = 20Mbps, and the total offered traffic is 10Mbps. The solid contours indicate
(C2 , ) parameter pairs with constant user throughput across the two cells (Mbps). The red dashed
line indicates optimum traffic split opt that maximizes (20).
12 J. Salo
ing is said about the load of the surrounding PRB utilization versus spectral efficiency
2.5
cells. This raises the question of how Tsch =
5.2. Assumption of Equal Sharing of PRBs where = 0.84. The mismatch is caused by
One of the assumptions listed in Section 2 is a small number of active UEs generating high
that the scheduler shares PRBs equally between load, in other words while 1, x does not
active UEs in time and frequency domain. If this tend to a high value (theoretically ), but in-
assumption does not hold the user throughput stead saturates at small value.
will not scale as 1/x where x is the number of The right-hand panel of Fig. 10 illustrates the
active UEs. For example, a scheduler that priori- more general formula (3) which does not rely on
tizes UEs having high good signal quality would the M/G/1 PS model and the underlying Pois-
not satisfy this requirement. Another commonly son assumption. The throughput is simply the
implemented scheduler type, the Proportionally cell capacity divided by the average of x, i.e., the
Fair scheduler has been shown to allocate, on number of active UEs. The agreement with the-
average, resources equally between users in [15]. ory is better than with M/G/1 PS, since no as-
sumptions are made regarding the distribution
5.3. M/G/1 PS modeling assumptions of x.
As with all models to obtain any usable re- To conclude this section, the extremely handy
sults some assumptions need to be made. In and insightful linear degradation of scheduled
the case of M/G/1 PS the assumption is that throughput (8) does not hold in cases where the
number of active UE arrivals per second is Pois- underlying statistical assumptions are violated.
son distributed, which effectively means that The more general formula (3) makes no assump-
as approaches 100%, also x increases sharply tions on distribution of the active UEs and is
which in turn makes user throughput approach therefore more robust.
zero. In fact if the Poisson assumption holds ex-
actly, the average active UEs is given by [7]
E[ x ] = , x 0.
1
6. CONCLUSION
This could be considered a reasonable assump-
tion, especially in mass events and cells dom- This white paper discussed mapping of LTE
inated by smart phones. The other extreme is user throughput to radio utilization. The aver-
a network (or a cell) where USB data modems age scheduled throughput, calculated with the
or wireless routers are the dominant UE type. 3GPP method, was shown to be fairly accu-
A single active UE downloading a very large rately predicted by the cell capacity divided by
file, or a wireless router shared by multiple the average number of active UEs. Adding the
subscribers, can cause PRB utilization of 100% usual assumption that user data flow arrivals
for long periods of time, and at the same time are Poisson distributed this result was expressed
hourly radio counters will show high scheduled in terms of Physical Resource Block utilization,
throughput per user. the outcome being the well-known M/G/1 Pro-
Empirical evidence from several networks cessor Sharing formula that predicts linear de-
suggests that cells that do not accord to C (1 ) crease of user throughput with PRB utilization.
frequently still present linear degradation in , Real-world measurement examples were inter-
but at a less steep rate. A typical example is spersed throughout the paper to illustrate the
shown in Fig. 10. The left-hand side panel applicability of the results. Extensions to some
shows scheduled throughput and the M/G/1 common LTE use cases were presented, and ap-
prediction, and clearly the actual throughput plication to load balancing between frequency
from counters has less steep decline with ; layers was discussed. A discussion on the un-
the best linear fit results in 1 degradation, derlying assumptions concluded the paper.
14 J. Salo
REFERENCES
PIMRC, 2006
15. J. M. Holzman, CDMA Forward Link Water-
filling Power Control, Proc. IEEE VTC, 2000