You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/261238989

Review of studies on end-to-end QoS in LTE networks

Conference Paper · October 2013


DOI: 10.1109/AEIT.2013.6666818

CITATIONS READS

22 5,207

2 authors, including:

Alessandro Vizzarri
University of Rome Tor Vergata
24 PUBLICATIONS 89 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

QoS-QoE estimation for 4G-5G wireless networks View project

Edge Cloud Computing in Telecommunications View project

All content following this page was uploaded by Alessandro Vizzarri on 13 August 2016.

The user has requested enhancement of the downloaded file.


Review of Studies on
End-to-End QoS in LTE Networks
Sonia Forconi Alessandro Vizzarri
Department of Electronic Engineering Department of Electronic Engineering
University of Rome Tor Vergata University of Rome Tor Vergata
Rome, Italy Rome, Italy
sonia.forconi@uniroma2.it alessandro.vizzarri@uniroma2.it

Abstract — The study of the end-to-end QoS (Quality of


Service) features of wireless networks may give to mobile
operators ways to manage the network infrastructures from a
II. Architecture
trustworthy and robustness perspective, to optimize the network
usage and guarantee desired quality of services to end user. In A. LTE architecture
this paper a review of studies on end-to-end QoS in LTE LTE network is based on architecture called EPS (Evolved
networks is presented. The most important features of the LTE Packet Switched System) shown in Figure 1 [1].
(Long Term Evolution) 3GPP Standard are its network
architecture, its ability to support the end-to-end QoS and its EPS is composed by two main subsystems:
ability to differentiate the user services. The wireless standard
which has immediately preceded LTE is the WiMAX Standard. • Radio Access Network: Evolved UTRAN (E-UTRAN)
Such a standard is based on the same radio transmission system • Core Network: Evolved Packet Core (EPC)
(OFDM). Therefore in the first part of the paper we compare the
QoS aspects of the LTE and WiMAX. In the second part of the
paper we focus on QoS aspects in E-UTRAN (Evolved Universal
Terrestrial Radio Access Network) section of LTE networks. In
the third part of the paper we also analyze the QoS aspects of the
end-to-end LTE network: the E-UTRAN section and the EPC
(Evolved Packet Core) section.

Keywords: LTE, QoS, QCI,

Fig. 1. LTE architecture (Font: 3GPP)


I. Introduction
Radio mobile technology has become very important in The main network elements of E-UTRAN are:
electronic market of the last years. Due to very large services • UE: User Equipment
they can deliver to end user, handheld terminals, such as • eNB: evolved Node B
mobile phones, notebook and pads, are now very popular in
our society. In fact we can use them for services like voice call UE is terminal of end user, while eNB is Radio Base
or internet connections, chat or messaging, activities in social Station of Mobile Network Operator (MNO) deputed to
networks. deliver radio signals.
The 4th Mobile Generation, called LTE (Long Term
Evolution), promises to deliver these services with very high The main network elements of EPC are:
performances, especially in terms of data rate: a speed of 300 • MME: Mobility Management Entity
Mbit/s per user is expected. • HSS: Home Subscriber Systems
Since LTE is the first FULL IP Wireless Technology • S-GW: Serving PDN (Packet Data Network) Gateway
standardized by 3GPP, it has a strong relationship with • P-GW: Proxy PDN (Packet Data Network) Gateway
applications based on data transfers such as internet • PCRF: Policy and Charging Rules Function
connections, downloads or FTP transfers, web browsing or
video streaming. MME is network node deputed to manage end user’s attach
procedures, mobility, handover and traffic balance operations.
HSS verifies end user subscription, S-GW is transmission
node linking E-UTRAN and EPC, P-GW is transmission node
linking LTE Core Network (EPC) and other networks.
PCRF is network node able to define QoS (Quality of Service)
policies of a particular service requested by user, so its
presence is crucial for end to end QoS approach.

B. LTE QoS Key Factors


Mechanisms of QoS management allow the mobile operator to
enable the privileged treatment and priorities for certain users
by the service differentiation. Bearer [2] is a fundamental
element related to QoS in LTE Network. Fig 2: QCI table (Font: 3GPP)
Bearer is a virtual path that delivers IP packet flows from the
terminal (UE) to the PDN gateway (in Uplink) and viceversa • ARP (Allocation Retention Priority): parameter used
(in Downlink). Bearer can transports the IP data flows with the for the management of priority in access procedures
same QoS requirement. in case of cell congestion. Allocation and retention
priority method is used to decide whether new bearer
Bearer can be classified in the following types: modification or establishment request should be
accepted considering the current resource situation
• GBR (Guaranteed Bit Rate): bearer for the transport
of the IP packet flows with high priority (e.g. VoIP • MBR / AMBR (Maximum Bit Rate / Aggregate
packets); it requires dedicated resources for the Maximum Bit Rate): Parameter used to specify the
duration of the transmission and high data rate, delay maximum bit rate supported by a bearer.
and low error rates.

• Non-GBR (Non-Guaranteed Bit Rate): bearer cannot


guarantee the availability of the resources allocated III. QoS in LTE
during the entire transmission, so it is established for This section of paper underlines different approach in
applications that do not require particularly high bit order to study end to end QoS in LTE, starting from the first
rate. documents published during LTE standardization process
(2008) and then arriving to the recent studies focused on QoS
• Default bearer: bearer established during the UE analysis in entire LTE transmission chain.
attach procedure to the mobile network; it provides
basic connectivity and it is suitable for transport of A. WiMAX vs LTE Qos
Best Effort traffic. The guarantee of QoS is an essential aspect for provision of
Internet services in 4G broadband wireless network and its
• Dedicated bearer: bearer established after the default main purpose is to meet the quality perceived by the user and
bearer if the IP packet requires specific QoS then to make optimal management of network resources.
parameters and therefore a specific QoS treatment
As first approach, QoS concept is studied for both WiMAX
QoS in LTE is characterized by the following parameters: and LTE networks making a comparison based on common
features OFDMA transmission scheme. Theremore a
• QCI (Quality of Service Class Identifier) is a scalar dedicated QoS framework is developed by Alasti, Neekzad,
used to manage functions (scheduling, queue Hui and Vannithamby et al [3].
management), relating to the forwarding of packets In [4] Bhagat, Halgaonkar and Wadhai present LTE and
on a network. Each QCI is characterized by priority WiMAX as two wireless broadband technologies designed to
level, packet delay budget and acceptable packet loss support QoS. So, LTE has a faster and simpler deployment
rate. The set of standardized QCIs are shown in mechanisms than WiMAX and it can provide better
Figure 2 [2]. Each bearer has an associated QCI performances in terms of reduction of latency, high data rate
value. and rapid HARQs.

Four main key factors must be considered for an useful


comparison of QoS techniques in WiMAX and LTE:

a. WiMAX is a standard IP wireless (IEEE 802.16) system


and it defines IP network architecture for the air access
interface, composed by Terminal equipments, Subscriber A set of performance indicators (throughput, delay, packet
stations and Base Station. LTE standard defines a FULL loss rate) are evaluated layer-by-layer. The proposed model is
IP architecture composed by the air interface (E-UTRAN) streaming service-oriented and the scenario consists of mobile
and a core network (EPC). The first studies of QoS in LTE terminals connected to a streaming server over LTE
take inspiration from those related to WiMAX network air technology.
interface.
In [6] impact of QoS support in a LTE network is studied in
b. Both LTE and WiMAX air interfaces use the physical case of delivering multiple services. For this purpose, in [6] a
layer radio access technology called OFDMA (Orthogonal system simulation is presented in order to assess network
Frequency Division Multiple Access) that is a digital capacity in two downlink scenarios. The first one is based on
multi-carrier modulation scheme. best effort network without QoS provisioning, while in the
second one on LTE network with QoS provisioning.
c. The QoS transport unit in WiMAX is called IP Service Both of them define capacity as the maximum number of users
Flow and it represents unidirectional flow of packets in in the cell and perform simulations with user services like
uplink or downlink direction. The LTE QoS transport unit VoIP, Real-Time Video, Mobile TV and Web Surfing.
is the EPS bearer, a virtual path between UE and PDN- Analytical model for the best effort network without service
GW. differentiation is presented and simulated with generated
packets placed in their arrival order. The other analytical
d. In WiMAX Standard Service Flows are treated following model is presented and simulated for the scenario with two
different criteria defined by 5 QoS service classes (UGS, queues of generated packets: their allocation depends on type
ertPS, rtPS, nrtPS, and BE). In LTE standard two types of of generating services.
bearer are considered: GBR bearer, in order to guarantee a Traffic mix combination both for the best effort network and
minimum data rate like rtPS), and non-GBR Bearer, used for network with QoS provisioning are simulated to measure
for best effort traffic. the cell capacity. Simulation results of two simulated scenarios
are compared. Importance of techniques for service
This four key factors show a strong link between approach differentiation in case of delay-critical services (e.g. VoIP) is
to QoS in WiMAX and LTE. They also underline LTE’s demonstrated, especially if they are delivered in combination
necessity to make an efficient tradeoff between high with delay-insensitive traffic services like TCP or Web
performances (especially in terms of data rate) and QoS Browsing. Conclusions underline benefit of QoS management
guaranteed for end user. able to support mobile operator for service differentiation and
prioritization.
B. Qos in E-UTRAN
Papers published after LTE vs WiMAX comparison focus The paper [7] proposes a combined Admission Control (AC)
their attention on radio access network (E-UTRAN) of LTE. and Packet Scheduling (PS) framework for QoS provisioning
The main concept of [5] is a LTE QoS model valid for in LTE uplink transmissions. To provide QoS control both AC
evolved 3G radio interface (E-UTRAN) and able to estimate and PS systems must be QoS-aware. The QoS aware AC
radio link performance between the UE and the eNodeB. This algorithm can decide if admit a new user in a cell depending
can make an efficient provision of a service. on cell traffic load information. This is realized using
The general scenario is based on downlink performance Fractional Power Control (FPC) algorithm that checks if
measurement at different protocol layers: PHYsical (PHY), Physical resource Blocks (PRBs) required from the new user
Medium Access Control (MAC), Radio Link Control (RLC) are available and if they can be assigned to him. The QoS
and Packet Data Convergence Protocol (PDCP). aware PS allocates dynamically shared data channels to the
QoS Model at PHY layer gives the configuration to a variable- active radio bearer by two steps. The first time-domain packet
rate multiuser and multichannel scenario: this gives channel scheduling (TD-PS) is used to set users’ prioritization. User’s
capacity in a typical LTE network. At MAC layer the model multiplexing is made using frequency-domain packet
defines two different QoS-scheduling algorithms: Round scheduling (FD-PS).
Robin (RR) and Modified Largest Weighted Delay First (M- FD-PS algorithm can adopt two different metrics. The first
LWDF). Simulations at PHY and MAC layers give indications one is a Proportional Fair Scheduled FD-PS metric built on
about the achievable users’ throughput. Simulation results relationship between the achievable throughput of user on
show that user throughput depends not only on the channel physical resource block and average of scheduled user
quality and relatives load conditions but also by the specific throughput. The second metric is GBR-aware FD-PS that
multiplexed scheduling algorithm. gives higher priority to a user which is far below its GBR
The proposed QoS Model follows a cross-layer mechanism for requirements.
the protocol stack of the radio link and it shows the QoS Performances are estimated using different Key Performance
needed at different layers. The approach is bottom-up and it is Indicators (KPI), such as average user throughput, average cell
based on cumulative performance degradation at different throughput, blocking probability and outage probability. As in
protocol layers of E-UTRAN (PHY, MAC, RLC e PDCP). paper [6], this document underlines importance of techniques
for QoS provisioning and service differentiation in a mixed important section of entire LTE mobile network because it is
traffic scenario, especially in entire LTE network (both directly linked to end user. Thus transmissions over radio
network access and core network subsystems). waves must follow particular criteria, especially in terms of
Paper [8] is a recent study on LTE network performance coverage and capacity.
evaluation of cross layer QoS scheduling. LTE software As discussed in [10], a particular E-UTRAN subsystem -
simulation tool is used to analyze performance of Cross Layer named MAC Scheduler - is dedicated to allocate radio
Two-way Scheduling Algorithm (CTSA) for treatment of Real resources according to the type of service to deliver (voice call
Time (RT) and Non Real Time (NRT) services in a multi-cell over IP, web browsing or video streaming).
scenario environment. The CTSA algorithm is developed The goal of this paper is to study how MNO must deliver
using different types of scheduler, like Proportional Fair (PF), VoIP, Video, HTTP and FTP services to end user over LTE
Modified Long Weighted Delay First (M-LWDF) and networks considering different QoS requirements of each
Exponential Proportional Fair (EXP/PF) scheduler. The PF bearer, firstly GBR (Guaranteed Bit Rate) and secondly non-
scheduler aim is to maximize the total network throughput and GBR.
to guarantee fairness among traffic flows. The M-LWDF A model of LTE MAC Scheduler with multi QoS aware
scheduler erases the RT packets from MAC queue if they are functions is presented using two different approaches: Time
not transmitted before expiration of their fixed deadline. The Domain (TD) scheduling module (used to differentiate the
EXP/PF scheduler increases the priority of RT flows. Scenario users according to their QoS characteristics) and Frequency
considered for simulations is multi cell and multi users one. Domain (FD) scheduling module, responsible of radio
Most important results of simulations are given in terms of resources assignment procedures (i.e. PRBs, Physical
throughput, packet loss ratio, delay and cell spectral Resource Blocks) among different users’ priority.
efficiency, and fairness index. They are strictly related to a FD scheduler defines a Priority Factor (PF) as ratio between
cross layer approach only used in Radio Access Network weight of MAC QoS class and data rate accumulated in the
Subsystem (E-UTRAN). previous steps. TD Scheduler analyses GBR candidate list
provided by the TD scheduler. Thus PRBs (Physical Resource
C. End to end Qos Blocks) allocation process is iterative, where one PRB
And end-to-end approach of a QoS guaranteed in 3GPP associated to the highest SINR value is allocated for one
mobile networks was already studied by 3GPP Work Groups. bearer during each iteration.
In fact [9] presents guidelines to be followed by Mobile Therefore a simulation activity is conducted considering 3
Network Operators (MNO) in order to assure an efficient end- different scenarios, four type of QoS services (VoIP, Video,
to-end QoS treatment in UMTS networks. HTTP and FTP), their mix usage and their delivering with
different priority.

Fig 3: Simulation Scenarios [10]

The objective of simulation is to study LTE network


performance in case of delivering four mixed services within
one QoS class, firstly mixing GBR and non-GBR ones and
secondly separating them.
Fig 3: End-to-End QoS Architecture (Font: 3GPP) So in the Scenario 1 all four services are delivered to users
(with no GBR prioritization), in Scenario 2 VOIP and Video
In this 3GPP Technical Specification end-to-end QoS bearers (GBR) are separated from HTTP and FTP bearers
management architecture is described, focusing on bearer (non-GBR), in Scenario 3 each service is delivered by one
services, integration with IMS systems and multimedia different bearer (in this case there are two GBR bearers and
services and relationship with network control plane functions. two non-GBR bearers).
So minimum levels can be defined for a QoS of a particular Simulation results show performances of VoIP bearers are still
service in terms of Bandwidth, Delay, Packet Loss Rate (PLR) good even when they are mixed delivered with the non-GBR
e Jitter. bearers in one MAC QoS class. Thus is true because VoIP
In LTE networks a strong interest for end to end QoS is still bearers have relatively small data rate which inherently gave
growing up, but the first papers analyze only Radio Interface them higher priority.
Access Network (E-UTRAN). E-UTRAN is the most In contrast, the video bearers suffer significantly when mixed
with the non-GBR bearers because of their higher data rate.
Since HTTP bearers deliver a lower data volume than FTP mobility. In this case end to end delay in the entire
ones, a good practice is to deliver FTP with no mixing it in transmission chain begin to affect considerably service
one MAC QoS class and then to assign the HTTP on a higher transmissions, VoIP in particular.
MAC QoS class than FTP. [12] underlines importance of simulation activity and End-to-
End (E2E) delay measurement in real-time applications (e.g.
Approach of [10] is mostly focused on LTE services QoS VoIP) to be performed in order to study overall QoS
considering user plane, but a complete approach to LTE performance in LTE networks.
performance estimation may also consider correlation between This paper present a simulation activity conducted considering
user plane and control plane [11]. Thus can be provided by three network scenarios: Baseline VoIP network with 50%
interaction of LTE backhaul network with LTE core network. network load, VoIP network with 95% network load and VoIP
Since a powerful end-to-end Performance Management is congested with FTP network scenarios.
strictly related to a good QoS perceived by end user, LTE For each scenario four cases are evaluated, one case with
backhaul network allows MNO to consider factors of real stationary node and other three cases with mobile nodes
conditions of their networks (data traffic, congestions, faults) (gradually increasing the node speed). The simulation results
and then to prevent negative effects of service delivering. for baseline VoIP network (non-congested), congested VoIP
Relationship between user plane and control plane is described network and congested VoIP with FTP network show that as
by an end to end bearer QoS model using an efficient the speed of node is gradually increased, E2E delay is slightly
CoS/QoS mapping using DSCP. So a final table with Data increased.
Path (User Plane) and Signaling Path (Control Plane)
informations can be defined.
Main criteria of CoS/QoS mapping involve two dimensions:
cross-layer mapping and cross-domain mapping. In this way
QoS mechanisms are used both in each layer of the protocol
stack and in each portion of the network, addressing great IV. Conclusions
benefits in terms of network throughput and service
Authors analyzed different approach to guarantee QoS in
availability. LTE networks. After WiMAX development, the first studies
3GPP Technical Specifications specify LTE QoS mechanism on LTE network focused on the performance evaluation of
defining QoS Class Identifier (QCI) and linking it to DSCP QoS-scheduling techniques at the radio interface. It means a
information. This mechanism is only valid for LTE networks performance evaluation between User Equipments (UE) and
domains with the same DSCP values. the eNodeB (eNB). Importance of QoS provisioning system is
Since transport networks must carry out IP Packet coming demonstrated by models and frameworks based on packet
from heterogeneous environment (not only LTE networks), scheduling algorithms, cross layer mechanism, service
may be useful mapping DSCP to QoS classes in transport differentiation focused on radio access. Only recent studies are
networks following this approach: focusing on end to end QoS approach for LTE, because
- Each EPS bearer is associated with a QCI and ARP Mobile Network Operators (MNO) are forced to make a
(Allocation and Retention Priority) information tradeoff between high performance of this new radio mobile
- LTE network equipment will map QCI to DSCP before technology (in terms of network performance, both access and
forwarding traffic towards transport network. Each DSCP transport networks) and QoS perceived by end user. This issue
identifies a unique QCI and DSCP markings are written to is becoming very crucial for LTE full deployment and roll out.
the ToS field in the header of IP packets Next works will study this end to end approach and they
- Transport network should apply DSCP markings in the will define a framework to be considered for guaranteeing a
traffic in order to implement functions of classification, good service quality to end user.
prioritization, and schedule.
This workflow is realized by apposite Performance Simulations of this approach will be provided.
Management System (PMS) composed by several logic nodes.
The most important one is QoS Diagnostic Engine, because it
can determine whether the network is impaired (the QoS
cannot be guaranteed) and data correlation from other logical Acknowledgements
modules.
In this way Performance Monitoring System (PMS) is able to The authors gratefully acknowledge the contributions of Prof.
compose CoS/QoS mapping table and then to detect possible Giuseppe Iazeolla (University of Rome Tor Vergata) and
impairments in transport networks given by mismatch of CoS Maurizio Siviero, Ivano Guardini and Roberto Procopio
and QoS mapping. (Telecom Italia) for their important contributions on final
In order to make an efficient study of end to end LTE QoS not version of this document.
only transport networks are to be analyzed but also service
reception conditions of end user [12]. A good perceived
quality is mandatory especially in condition of end user
References

[1] 3GPP TS 36-Series, http://www.3gpp.org/.


[2] 3GPP Technical Specification 23.207, “End-to-end Quality of Service
(QoS) concept and architecture”, http://www.3gpp.org/ftp/Specs/html-
info/23207.htm.
[3] M. Alasti, B. Neekzad, C.J. Hui and R. Vannithamby, “Quality of
Service in WiMAX and LTE Networks”, Intel Labs IEEE
Communications Magazine, , vol. 8, pp. 104-111, May 2010.
[4] P.M. Bhagat, P.S.Halgaonkar, and V.M. Wadhai, “Comparison of LTE
and WiMAX on the Basis of Qualities”, International Journal of P2P
Network Trends and Technology, vol. 1, 2011.
[5] G. Gomez, J. Poncela González, M. C. Aguayo-Torres, J. F. Paris, J. T.
Entrambasaguas. “QoS modeling for performance evaluation over
evolved 3G networks”, in Proc, of the 3rd ACM Workshop on QoS and
Security for Wireless and Mobile Networks (Q2SWinet), Oct. 2007.
[6] I. Siomina and S. Wanstedt, “The Impact of QoS Support on the End
User Satisfaction in LTE Networks with Mixed Traffic”, Personal,
Indoor and Mobile Radio Communications, 2008. PIMRC 2008. IEEE
19th International Symposium on, pp. 1 – 5, 15-18 Sept. 2008.
[7] M. Anas, C. Rosa, F.D. Calabrese, K.I. Pedersen, and P.E. Mogensen,
“Combined Admission Control and Scheduling for QoS Differentiation
in LTE Uplink”, Vehicular Technology Conference, 2008. VTC 2008-
Fall. IEEE 68th, pp. 1 – 5, 21-24 Sept. 2008.
[8] D. Vinayagam, R. Kurinjimalar, D. Srinivasan, ” Performance
evaluation of cross layer QoS scheduling for Long Term Evolution
Network”, International Journal of Advanced Computer Research, vol. 2
number 3 issue 5, September 2012.
[9] 3GPP TS 23.207 v6.6.0, “3rd Generation Partnership Project; Technical
Specification Group Services and System Aspects; End-to-end Quality
of Service (QoS) concept and architecture”.
[10] Y. Zaki1, T. Weerawardane1, C. Görg1, and A. Timm-Giel, “Multi-
QoS-Aware Fair Scheduling for LTE”, Vehicular Technology
Conference (VTC Spring), 2011 IEEE 73rd , pp. 1 – 5, 15-18 May 2011.
[11] L. Li and S. Shen, “End-to-End QoS performance management across
LTE networks”, Network Operations and Management Symposium
(APNOMS), 2011 13th Asia-Pacific, pp. 1 – 4, 21-23 Sept. 2011.
[12] M.T. Aziz, M.E. Masum, M.J. Babu, S. Rahman, and J.Nordberg,
“Mobility Impact on the End-to-End Delay Performance for VoIP over
LTE”, International Conference on Communication Technology and
System Design 2011, vol. 30, 2012, pp 491–498.

View publication stats

You might also like