Sei sulla pagina 1di 5

Cell-Specific Uplink Power Control

for Heterogeneous Networks in LTE


Jacek Góra1), Klaus I. Pedersen2), Agnieszka Szufarska1), Stanisáaw StrzyĪ1)
1) 2)
Nokia Siemens Networks, Wroclaw Nokia Siemens Networks, Aalborg
E-mail: {jacek.gora, klaus.pedersen, agnieszka.szufarska, stanislaw.strzyz}@nsn.com

Abstract— This paper provides an assessment of 3GPP coordination schemes [1], various femto access strategies [2],
standardized LTE/LTE-A uplink power control procedures in off-load and architecture aspects [3], etc. Majority of analysis
case of a co-channel operation of different types of base stations. for such cases focuses mainly on the downlink performance.
Especially the case of femtocell deployment with an overlaying
macrocell is considered. Imperfections of the currently accepted In this paper we investigate the performance of LTE uplink
procedures in case of such networks are pointed and a modified for macro- and femtocell co-channel deployment. We
solution is proposed. It is shown that in case of heterogeneous investigate if co-channel deployment is a feasible
networks more cell-specific configuration can lead to 8% and configuration, and if introducing HeNBs is possible without
14% performance improvements in capacity and coverage of degradation of the macrocell performance.
local cells respectively, without harm for overlying macrocells.
The evaluation of the proposed scheme has been done on case of LTE uplink performance is known to be sensitive to power
a macro and femto co-channel deployment, but should be also control settings. This has been proven for many macro-only
valid for other cell types. studies of uplink power control [4-9]. In this paper, we extend
these previous studies to investigate uplink power control
Keywords-LTE, uplink, power control, RRM, femto (ULPC) optimization and performance for co-channel
operation of heterogeneous networks.
I. INTRODUCTION The article is organized as follows: in section II the current
Local (low power) cells are foreseen as a solution for high state of the art approach standardized by the 3GPP organization
performance mobile networks. Shortening the distance between is presented. In section III uplink interference effects present in
a base station and user equipment (UE), and especially a heterogeneous network are discussed and a modified ULPC
bringing the access node indoors, should provide more procedure is proposed for local area cells. Section IV describes
effective power and spectrum utilization than in case of wide- the simulation model and procedures utilized for the presented
area macrocells. Networks based on a combination of analysis. Sections V and VI evaluate respectively the ULPC
macrocells and small cells are therefore envisioned as one of procedures currently existing in 3GPP standards and the
the most promising solutions for meeting the increasing traffic proposed modified ULPC scheme. Finally, section VII
demands. summarizes the main conclusions from the study.
The macrocells, however, cannot be abandoned completely.
II. 3GPP STANDARDIZED LTE UL POWER CONTROL
They are required to provide a continuous coverage between
small cells and in proximity of restricted access cells (e.g. due The uplink power control for LTE is defined by 3GPP in
to Closed Subscriber Group, CSG). A network composed of [10] as combination of open and closed loop components. The
different types of cells is commonly known in the 3GPP open loop power control (OLPC) is responsible for a rough
nomenclature as a heterogeneous network. setting of UE transmit power. It compensates slow changes of
pathloss (including shadowing) in order to achieve a certain
A type of small cell base station is a femtocell, in mean received signal power for all users. The closed loop
LTE/LTE-A also known as a Home eNodeB (HeNB). It is power control (CLPC) is used for user specific adjustments of
defined as a low power access point that uses licensed the power settings. As an example, CLPC can be used for
spectrum to communicate with users and a high-speed fixed further optimization of general network performance as
Internet connection (e.g. DSL) as backhaul towards the described in [8-9].
operator’s core network. HeNBs have the ability to support all
the devices and at least all the services that work in the wide The equation defined in 3GPP for setting transmit power of
area network. Their deployment in general is considered as Physical Uplink Shared Channel (PUSCH) is as follows
uncoordinated, i.e. without a priori radio network planning. (dB scale) [10]:
Although the research and standardization process of
femtocells is still ongoing, some operators have already started
P = min{ PMax, 10log10( M ) + P0 + Į · PL + ǻMCS + f(ǻi) } (1)
to introduce them in their networks.
There have been many studies on macro+femto networks. where PMax is the maximum allowed UE transmit power, M is
Many of those studies have investigated interference the number of Physical Resource Blocks (PRB) scheduled for

978-1-4244-3574-6/10/$25.00 ©2010 IEEE


the given user in a time slot, P0 is parameter related to target configuration parameters that are assumed to be operator or
mean received power (user or cell specific), Į is pathloss vendor specific. Finally, the function round( ) in (1) quantizes
compensation factor (cell specific), PL is the downlink pathloss the argument to the closest allowed parameter setting of P0 as
measured by the UE [11], ǻMCS is a parameter depending on allowed by the 3GPP standard.
the Modulation and Coding Scheme (MCS) used by the user,
and f(ǻi) is a user specific CLPC correction. The proposed method is a tradeoff between improving user
performance and simplifying network configuration. The
In this contribution the cell specific parameters of the algorithm introduces two additional parameters (APo, BPo are
OLPC are considered (P0, Į) as they have the main impact on the same for all HeNBs) and with it allows full individual
the inter-cell interference. As specified in the 3GPP standards configuration. To achieve similar effect with the basic OLPC
the OLPC parameters can have only specific values [13]: procedure, each local cell would have to be configured with an
individual set of parameters.
• P0: integer{ -126..24 } dBm
• Į: { 0.4, 0.5, 0.6, 0.7, 0.9, 1.0 }
IV. SIMULATIONS
The maximum (unrestricted) UL transmit power is 23 dBm and
This section presents the methodology used for evaluation
a minimum power a UE can use in connected mode for all
of uplink power control algorithms in case of macro+femto
active channels is -40 dBm [12].
networks. First the simulation model and later the performance
metrics and the assessment procedure are discussed.
III. POWER CONTROL IN HETEROGENEOUS NETWORKS
With uplink transmissions in heterogeneous networks, two A. Scenario and simulation model
cases should be distinguished. At the edge of a macrocell very The simulation methodology used to generate the results
high interference levels exist (HeNB #2 at Figure 1), because presented in this paper is inline with the 3GPP recommended
macro-UEs transmit with high powers to overcome high approach for HeNB evaluation [15]. In this paper a sub-urban
pathloss towards their serving base stations. In this case femto- (residential) scenario is considered (Figure 2). In this
UEs are the main victims. On the other hand, close to a macro- deployment type simple 12x12 m, single floor buildings are
eNB, the femtoĺmacro interference may become a serious modeled. Penetration loss of the buildings’ external walls is set
threat (HeNB #1 at Figure 1). to 10 dB. Internal walls have not been modeled explicitly.
Wanted signal
Interference

Macro-eNB HeNB #1 HeNB #2


Figure 1. Interference paths in uplink

As the level of interference measured by HeNBs in the


uplink is usually higher at the edge of a macrocell, femto-UEs
in such locations should use higher power to overcome the
interference. On the other hand, femto-UEs close to a macro-
eNB should use lower power settings in order not to generate Figure 2. Network deployment scheme
too much interference towards the macrocell. Satisfying needs
of the two cases requires a different set of OLPC parameters The most essential parameters of the model are summarized
depending on location of the local node in relation to in TABLE I. For additional information, see [15]. The SINR to
macrocell. Another solution is to define a procedure for self throughput mapping is based on a modified Shannon formula
configuration of local cells. The self configuration could be [16], adjusted to approximately follow the LTE link
done based on e.g. measurements of the macrocell signals done performance.
by the local node itself [14].
TABLE I. SIMULATION PARAMETERS (SEE ALSO[14])
The solution proposed in this study is to configure the
OLPC parameter P0 for the femto-UEs in a way that would Parameter Setting
reflect the distribution of interference levels within the
macrocell (adaptive P0) as a function of pathloss towards System configuration LTE, 10 MHz bandwidth (50 PRBs),
closest macro-eNB: full co-channel operation of macro- and femtocells,
4 PRBs reserved for PUCCH
P0 = round( APo + BPo · PLLA_WA ) (2) Transmit powers:
where P0 is the calculated OLPC parameter for a local area eNB 46 dBm
HeNB Controled in range of 0..20dBm as a function of
node, PLLA_WA is the downlink pathloss to the strongest co- signal power from strongest macro eNB :
channel deployed macrocell node. APo and BPo are PHeNB = max( 0, min( 20, PRx_eNB + 60dB )) [dBm]
Antenna configuration:
eNB 2 Rx, 2 Tx, 3 sectors, 2D patern, 14 dBi gain
HeNB 2 Rx, 2 Tx, omnidirectional, 5 dBi gain
UE 2 Rx, 1 Tx, omnidirectional, 0 dBi gain
Macrocell deployment Hexagonal, 500 m inter-site distance,
21 sectors simulated, statistics gathered over 3
central sectors
Femtocell deployment 10 HeNBs per sector droped randomly within
simple residential buildings (12x12 m)
User deployment 10 UEs per sector with access to eNBs only
1 additional UE for every deployed HeNB
Scheduler Round robin
Figure 3. Macrocell performance in case of “macro only” network
Trafic model Full buffer
Those settings are here-after used for OLPC of macro-UEs in
Shadowing: Simulated as in [15]
macrocell 50m correlation distance and 8dB standard
all investigated macro+femto scenarios. The results achieved
deviation for the “macro only” case with the selected configuration are:
femtocell 3m correlatio distance and 4dB standard deviation
• Average cell capacity: 12.20 Mbps
Fast fading Not simulated • Average cell coverage: 0.440 Mbps
• Average IoT at eNB: 22.68 dB
The serving cell selection procedure for the UEs is based
on Reference Signals’ Received Power (RSRP) measurements. V. BASIC POWER CONTROL AT LOCAL NODES
UEs connect to the cell that provides the highest RSRP value at
The next set of results assumes the presence of both macro
a given location. For HeNBs a restricted access is assumed.
and femtocells. The aim of this analysis is to determine if using
Only the UEs that are part of the HeNB’s Closed Subscriber
only one set of OLPC parameters is attractive. Two cases have
Group (CSG) can connect and be served by it.
been analyzed. First when all base stations (macro and femto)
use the same OLPC settings and a second one when femtocells
B. Evaluation procedure use a different set of parameters than macrocells, but the same
The performance of ULPC procedures has been quantified one for all femtos.
based on several performance metrics. The main performance
metrics are: As illustrated in Figure 4 UEs connected to femtocells use
much lower transmit power than users served by macrocell.
Average cell capacity – aggregated throughput from all users This is because the pathloss in small cells is lower as compared
connected to a single cell, averaged over time [Mbps] to values observed in macrocells. This is an obvious advantage
which impacts directly the lifetime of a mobile battery.
Average cell coverage – lowest user throughput that can be
guaranteed in 95% of cases (5%-ile CDF) [Mbps]
Interference over Thermal Noise (IoT) – level of total
interference and thermal noise above the thermal
noise measured at a base station receiver [dB]
In case of heterogeneous networks the OLPC parameters
can be tuned to achieve better macrocell performance at the
cost of worse performance at femtocells and vice versa. An
appropriate configuration should be selected as a tradeoff
between performances of the two network layers. In this study
our objective is to set the OLPC parameters to avoid harming
the macro-UE performance as result of introducing femtocells.
In this sense, the optimal configuration is the one that provides
the best performance at femtocells, while at the same an
undisturbed macrocell performance (compared to a “macro
only” reference case).
The simulation of a “macro only” reference case was done
with all HeNBs turned off (all users left in their positions). Figure 4. Transmited (a) and received (b) power statistics in case of a
Based on the presented results in Figure 3, the OLPC macro+femto network with a basic OLPC procedure
parameter setting giving the most attractive performance is:
Lower transmit power influences as well positively the
• P0 = -80 dBm levels of IoT at the nearby base stations. If femto-UEs used
• Į = 0.9 similar power levels as macro-UEs, the general network
performance would degrade significantly.
The analysis of network performance with common and VI. ADAPTIVE P0 POWER CONTROL PROCEDURE
different uplink power control settings for macro- and For the cases discussed in Section V it has been observed
femtocells shows that having the same parameters for the that the level of uplink interference measured at the HeNBs
OLPC algorithm is suboptimal. As shown in Figure 5 in case varies significantly depending on the HeNBs location. There is
of common OLPC parameters, the femto-UEs are not able to a correlation between the pathloss from the HeNB to the
reach the maximum throughputs. closest macro-eNB and the measured level of IoT as is
In macrocells good pathloss compensation is very illustrated in Figure 6.
important (high Į value), whereas in small ones (femto) more
important parameter is the power offset (high P0 value). Having
the same OLPC settings for macro- and femtocells results in
very low transmit powers. Thanks to femtocell isolation in
form of building’s wall attenuation, femto-UEs are able to
increase transmit powers without harm to macro-UEs. This can
be achieved with different OLPC setting than used at
macrocells. .

Figure 6. Distribution of uplink interference in a macrocell

Different combinations of APo and BPo have been tested and


the optimal one has been selected. For that a procedure as
described in section IV B has been used. The combination of
settings APo = -84 dBm and BPo = 0.3 have proven to provide
the best performance of femto-UEs, without harming macro-
UEs. Figure 7 presents a distribution of P0 values calculated
with the parameters. In case of 90% femtocells the calculated
P0 setting is higher than the fixed P0 used with the basic OLPC
procedure for the same Į value. This means that the 90% femto
cells are able to provide improved radio conditions in uplink.

Figure 5. Macro (a) and femto (b) users’ UL throughputs in case of a


macro+femto nework with a basic OLPC procedure

With a different set of OLPC settings for femtocells (lower


Į and higher P0 than in case of macrocells), users with
generally very low pathloss values can achieve much better
SINRs (and available throughputs) without causing additional
harm to macrocell performance as shown in Figure 5. Table II
contains a summary of the achieved performance metrics for
the two discussed cases.

TABLE II. PERFORMANCE STATISTICS FOR A MACRO+FEMTO NETWORK Figure 7. Distribution of P0 parameter values used by local area nodes
WITH A BASIC OLPC PROCEDURE

Common OLPC settings for Different OLPC settings for Figure 8 shows the user throughput statistics with the basic
all cells (macro and femto) macro- and femtocells OLPC and adaptive P0 power control procedures. Even though
Macrocell Femtocell Macrocell Femtocell in most of the cases, femto-UEs use higher transmit powers
P0 = -80 dBm P0 = -80 dBm P0 = -80 dBm P0 = -59 dBm with the adaptive P0 scheme than with the basic OLPC, the
Į = 0.9 Į = 0.9 ǹ = 0.9 Į = 0.6
performance of macrocell users does not change significantly.
Cell
capacity 12.32 26.09
12.22 34.19 At the same time femtocell performance is improved. The
(-1%)1 (+31%)1 average cell capacity and the coverage is improved by 8% and
[Mbps]
Cell 14%, respectively (see details in Table III). More than twice as
0.707 24.25
coverage 0.712 18.63
(-1%)1 (+30%)1
many femto-UEs are able to achieve top transmission speeds
[Mbps] and usually a 4 Mbps gain is possible in average user
Average
22.47 12.61 22.57 12.73 throughputs.
IoT [dB]
1
The percentage values show change related to the common OLPC settings case
small cells is the increase of mobile batteries lifetime, as the
transmit power used by the femto-UEs is much lower than the
macro-UEs.
It has been shown that a common set of ULPC parameters
for all cells is suboptimal in terms of available user
throughputs. It has, however, the benefit of the simplest
configuration (minimal number of parameters). Just by using a
separate sets of ULPC parameters for macro- and femtocells, it
is possible to increases average femtocell capacity and
coverage by approximately 30%, without decreasing
performance of macrocell users. The gain comes, however, for
the cost of more complicated network configuration.
Further optimization of femto user performance is possible
by using a cell-specific configuration for femtocells. An
algorithm has been proposed for an autonomous configuration
of ULPC parameters at small cells. With it an additional 8%
capacity and 14% coverage gains at femtocells are possible,
without harm for macrocell performance.
Figure 8. Macro (a) and femto (b) users’ UL throughputs in case of basic
OLPC and adaptive P0 procedures The presented study has been focused on restricted access
(CSG-type) femtocells. Future work on this subject should also
Figure 9. Performance statistics for a macro+femto network with the include open access small cell such as picocells for hot-zones,
adaptive P0 procedure and relaying nodes.
Basic OLPC Adaptive P0 power control
Femtocell REFERENCES
Macrocell Femtocell Macrocell
APo = -84 dBm [1] D.López-Pérez, et al., “OFDMA Femtocells: A Roadmap on
P0 = -80 dBm P0 = -59 dBm P0 = -80 dBm
BPo = 0.3 Interference Avoidance”, IEEE Communications Magazine,
Į = 0.9 Į = 0.6 Į = 0.9
Į = 0.6 January 2010
Cell [2] G.de la Roche, et al., “Access Control Mechanisms for Femtocells”,
12.20 37.04
capacity 12.22 34.19 IEEE Communications Magazine, January 2010
(-0%)1 (+8%)1
[Mbps] [3] D.Calin, et al., “On Femto Deployment Architectures and Macrocell
Cell Offloading Benefits in Joint Macro-Femto Deployments”, IEEE
0.706 27.62
coverage 0.707 24.25 Communications Magazine, January 2010
(-0%)1 (+14%)1
[Mbps]
[4] C.U.Castellanos, et al., “Performance of Uplink Fractional Power
Average Control in UTRAN LTE”, VTC 2008 Spring, Marina Bay, Singapore,
22.57 12.73 22.60 12.87
IoT [dB] May 2008
1
The percentage values show change related to the basic OLPC procedure [5] N.J.Quintero, “Advanced Power Control for UTRAN LTE Uplink”,
M.Sc. thesis, Aalborg University, June 2008
The gain from the proposed adaptive P0 procedure is [6] S.A.Elayoubi, O.B.Haddada, “Uplink Intercell Interference and Capacity
in 3G LTE Systems”, ICON 2007, Adelaide, Australia, November 2007
possible because UEs in the femtocells placed closest to a
[7] C.U.Castellanos, et al., “Uplink Interference Control in UTRAN LTE
macro-eNB have lowered transmit powers (low P0). Those are Based on the Overload Indicator”, VTC 2008 Fall, Calgary, Canada,
the UEs that are mainly contributing to interference at the September 2008
macrocells. At the same time, femto-UEs with a high pathloss [8] M.Boussif, et al., “Interference Based Power Control Performance in
towards a macro-eNB can use an increased transmit power LTE Uplink”, ISWCS’08, Reykjavik, Iceland, October 2008
(high P0), as they affect the macrocell performance on a much [9] M.Boussif, et al., “Load Adaptive Power Control in LTE Uplink”,
lower level. European Wireless 2010, Lucca, Italy, April 2010
[10] 3GPP, “E-UTRA – Physical layer procedures”, TS 36.213 v8.8.0
VII. CONCLUSIONS [11] 3GPP, “Physical layer aspects for evolved Universal Terrestrial Radio
Access (UTRA)”, TS 25.814 v7.1.0
In this paper the topic of uplink power control assessment [12] 3GPP, “E-UTRA – UE radio transmission and reception”, TS 36.101
and optimization in case of macro+femto co-channel operation v9.1.0
has been discussed. The conclusions that can be drawn from [13] 3GPP, “E-UTRA – MAC protocol specification”, TS 36.321 v9.0.0
the study are that with appropriate uplink power control [14] 3GPP, “E-UTRA – Physical layer - Measurements”, TS 36.214 v8.7.0
settings, introduction of femto nodes does not jeopardize the [15] 3GPP RAN4, “Simulation assumptions and parameters for FDD HeNB
performance of macro users compared to a macro only case. RF requrements”, R4-092042, May 2009
And typically the throughputs at femto cells are far better then [16] P.Mogensen, et al., “LTE Capacity Compared to the Shannon Bound”,
those available at macrocells. A secondary benefit of deploying VTC 2007 Spring, Dublin, Ireland, April 2007

Potrebbero piacerti anche