Professional Documents
Culture Documents
5G
5G
Performance Assessment of
Open Software Platforms for 5G Prototyping
Francesco Gringoli, Paul Patras, Carlos Donato, Pablo Serrano, and Yan Grunenberger
Digital Object Identifier: Francesco Gringoli is with the University of Brescia/CNIT; Paul Patras is with the University of Edinburgh; Carlos Donato is with the
10.1109/MWC.2018.1800049 University of Antwerpen — imec; Pablo Serrano is with University Carlos III of Madrid; Yan Grunenberger is with Telefonica I+D.
32
70 17.8 16.8 11.2 10.4 10.4 1.6 10.4 12 70 30.3 32.9 30 33 33.1 32.9 29.9
31.5 • Measuring the achievable throughput by running
75 17.8 17.8 16.7 11.1 10.9 10.9 10.9 10 75 32.7 32.8 33 29.2 33 32.9 33.1
31
five consecutive transmissions, each of 5 s duration
80 1.6 1.6 1.6 1.6 1.6 1.6 13.5
8
80 31.5 30.6 33.1 32.8 30.4 33.1 33.1
After the network has been set up successfully,
6
30.5
we run 30 s long transmission experiments, setting
85 1.6 1.6 17.8 1.6 1.6 1.6 1.6 85 32.8 32.9 33.1 33 32.9 32.9 32.3 30 the offered close to the the maximum throughput
4
90 1.6 1.6 1.6 1.6 1.6 1.6 1.6 90 33 32.8 32.8 33 32.7 33.1 32.8 29.5 measured at the last step of the bootstrap procedure.
2
60 65 70 75
RX gain [dB]
80 85 90 60 65 70 75 80
RX gain[dB]
85 90
Validation
To confirm that the proposed evaluation method-
FIGURE 1. Impact of eNB RX/TX gain on the throughput achievable in the ology is reliable and not susceptible to inaccuracies
uplink (left) and downlink (right) directions. The eNB employs the srsENB induced by practical signal attenuation and mul-
stack over a 10 MHz channel, the UE is a Huawei USB dongle, and the tipath propagation, we first compare the through-
propagation medium is coaxial cable. put attainable in both directions (UL and DL) over
5 and 10 MHz channels with all the possible soft-
ware/hardware configurations when the eNB and
eNB/UE combination OAI/Huawei OAI/srsUE SRS/Huawei SRS/srsUE UE communicate over a wireless channel (air) and
over coaxial cables using SMC connectors (cable),
respectively. Note that our validation does not
include experiments with the Nexus smartphone,
30 as instrumenting wired connections would have
voided the device’s warranty. We illustrate the
DL, 5MHz
results of these preliminary experiments in Fig. 2.
Throughput (air) [Mb/s]
BW = 10MHz
ferences between the commercial equipment’s per- 20
formance (i.e., Nexus and Huawei) and that of the
experimental platform (srsUE). In particular, for the 10
Throughput [Mb/s]
case of {DL, 10 MHz} (top left subplot) we note a
0
slight drop in performance when srsUE is employed
with OAI. We also note that initially the throughput
attained with the (SRS, Nexus) combination was 30
lower. Our intuition is that the signal processing per-
BW = 5MHz
formed by the SRS stack (in both the eNodeB and 20
BW = 10MHz
Extensibility and Pitfalls
50
Customization and Extensibility
We next comment on the ability to customize and
extend the functionality of the platforms consid-
CPU [%]
0
ered. To this end, we focus on particular issues
related to scheduling and modulation and cod-
100 ing scheme (MCS) assignment. To ensure that
BW = 5MHz
the two solutions studied are evaluated under
the same conditions and all comparisons are
50 fair, we decided to focus on the following cus-
tomization: introduce the ability to fix, during
experiments, the MCS assignments that the eNB
0
Nexus Huawei srsUE Nexus Huawei srsUE
MAC scheduler enforces on UEs.6 Achieving this
UE turned out to be fairly intuitive with srsLTE, as we
found the function responsible with scheduling
FIGURE 4. CPU usage at the eNB for different eNB stacks, UEs, bandwidths, and MCS assignment in srsLTE/srsenb/src/
and transmission directions. mac/scheduler_ue.cc, conveniently named
sched_ue, and the code was easy to modify.
this purpose, the central lines marking the medians, On the other hand, this task turned out to be
the boxes’ lower and upper margins the 25th and less straightforward with OAI. The source code
75th percentiles, respectively, and the whiskers the that implements the MCS assignment operation is
minimum and maximum values, excluding outliers, located within the folder openairinterface5g/
which we plot separately (crosses). openair2/LAYER2/MAC/, where files related to
Observe that the SRS stack displays the most MAC scheduling and MCS index assignation are
deterministic behavior, especially with the Huawei located. After thorough code inspection, we found
dongle and the SDR running srsUE. Indeed, if we that all the files therein contain code that chang-
exclude outliers, the bootstrap time is constant and es the MCS settings, and unfortunately, the MCS
less than 22 s. With the same UE types, the OAI index is also often hard coded in places. Following
stack performs substantially differently. In particular, debugging, we inferred that the files eNB_sched-
while the bootstrap time is fairly constant and close uler_ulsch.c and eNB_scheduler_dlsch.c
in magnitude to that observed when the eNB runs contain the functions schedule_ulsch_rnti
the SRS stack (approximately 25 s median value) and schedule_dlsch_rnti, which assign the
and the srsUE is used, the range of bootstrap times MCS in the UL and DL directions. We developed
is very large when the UE used is the USB dongle a patch to enable dynamic MCS index assignment,
(Huawei). In this case, the median is also higher although after applying our patch we discovered
than that measured in all the other setups, and the that the MCS will be later computed and altered
75th percentile is more than 70 s. This is particularly by other functions. Therefore, we were unable to
inconvenient since when experimenting for perfor- achieve the desired behavior.
mance assessment purposes, setting up the network We believe the major differences between the
takes more time than running the actual traffic. If the OAI and srsLTE solutions in terms of extensibility are
Nexus smartphone is used as UE, both eNB types because they follow different software designs. In
exhibit similar statistics, that is, the median of the particular, OAI was developed for mock LTE network
bootstrap time is approximately 28 s, and the varia- deployment with a built-in emulator, while srsLTE was
tions are relatively small (less than 2 s between the designed from scratch as a framework to support
1st and 3rd quartiles). We leave an analysis of the building LTE applications on top, providing a set of
reasons behind these differences for future work. common libraries, tools, and examples for PHY layer
We also count the number of times we detected implementation and experimentation. As a result, UE
a failure of the bootstrap procedure or during exper- and eNB apps were implemented on top of these
iment runtime for each of the eNB/UE combinations libraries. From a software design perspective, srsLTE
considered, in the process of conducting a total of offers a modular framework that re-factors the code
80 successful experiments in each case, which we of common LTE functions for any application, while
reported earlier (20 repetitions for each direction OAI is designed to offer a standalone eNB solution.
and bandwidth setting). Interestingly, we find that
the most reliable UE type is the srsUE, as we never Software Stack Pitfalls
encountered any failures when connecting this to Working with open source cellular stacks has
either of the eNB types used, and the network never important benefits, including speed of deploy-
failed during the experiments. We observed a similar ment, availability of documentation, affordable
6 In the case of srsENB, the behavior with the USB dongle only when connecting cost, and ability to extend functionality. Unfortu-
software includes code to
fix the MCS index at startup to the OAI eNB stack, while with SRS we measured a nately, such solutions come with their own set of
through a configuration file, 5.9 percent failure rate. Here, failures always occurred issues, some of which are more difficult to spot
but our aim is to dynamically after the network was formed successfully, during the and can hinder the reproducibility of results. Here
change the MCS externally initial short tests that we run to assess the sustainable we highlight the main pitfalls we identified while
during execution time.
throughput (step 5 of the setup procedure described experimenting with the OAI and SRS tools.
Time [s]
only two bandwidth settings, 5 and 10 MHz.
Interconnection with CN: The srsENB implemen- 40
tation employs the same subnetwork for both user
plane (S1-U interface) and control plane (S1-C inter- 30
face). On the other hand, the OpenAir-CN can be
configured to use two different subnetworks in order
to distinguish between the two planes; if such config- 20
uration is enabled, the srsENB stack will not work.
Problematic Queue Management: We note 10
that sending traffic in the DL direction at a rate OAI/Huawei SRS/Huawei OAI/srsUE SRS/srsUE OAI/Nexus SRS/Nexus
which exceeds the maximum throughput support- eNB/UE combination
ed on the channel makes OAI crash. Following
FIGURE 5. Distribution of the network bootstrap times (until the UE can ping
code inspection, we find that the different threads
the CN over the air) with the different eNB/UE combinations considered.
composing the software do not implement any
packing dropping strategy at the queues/lists used
for communication, which leads to out-of-memory [9] X. Foukas, M. K. Marina, and K. Kontovasilis, “Orion: RAN Slic-
ing for a Flexible and Cost-Effective Multi-Service Mobile Net-
issues. We have fixed this bug and are proposing a work Architecture,” Proc. ACM MobiCom, 2017, pp. 127–40.
patch to the OAI developers community. [10] X. Foukas et al., “FlexRAN: A Flexible and Programmable
Platform for Software-Defined Radio Access Networks,”
Summary Proc. ACM CoNEXT, 2016.
[11] A. Virdis et al., “Performance Analysis of OpenAirInterface
This article reports a performance assessment of System Emulation,” Proc. PMECT, Rome, Italy, Aug. 2015.
the two most prevalent open software solutions [12] N. Nikaein, “Processing Radio Access Network Functions
for mobile network prototyping, namely, srsLTE in the Cloud: Critical Issues and Modeling,” Proc. Int’l. Wksp.
and OAI. We designed a methodology to charac- Mobile Cloud Computing and Services 2015, pp. 36–43.
[13] S. Bhaumik et al., “CloudIQ: A Framework for Processing
terize the performance of these stacks, quantifying Base Stations in a Data Center,” Proc. ACM MobiCom, 2012.
their differences in throughput and resource con-
sumption over a range of practical settings. Our Biographies
findings formalize “word of mouth” knowledge Francesco Gringoli [M’04, SM’17] received his Laurea degree
in telecommunications engineering in 1998 and his Ph.D.
among practitioners, and provide useful guidelines degree in information engineering in 2002. He is an associate
for deploying 5G testbeds with these tools. professor of telecommunications in the Department of Informa-
tion Engineering at the University of Brescia, Italy. His research
Acknowledgments interests span the fields of wireless networking, including securi-
ty, medium access control, and the physical layer.
The authors are grateful to the anonymous referees
for their valuable comments, which helped in improv- Paul Patras [M’11, SM’18] received M.Sc. (2008) and Ph.D. (2011)
ing the article. We also thank Iñaki Úcar for his help degrees from Universidad Carlos III de Madrid (UC3M). He is a lec-
setting up the testbed. The work of P. Serrano was turer and Chancellor’s Fellow in the School of Informatics at the Uni-
versity of Edinburgh, where he leads the Internet of Things Research
partially supported by the European Commission in Programme. Previously, he was a research fellow at the Hamilton Insti-
the framework of the H2020 5G-PPP 5G-MoNArch tute of the National University of Ireland Maynooth. His research inter-
project (grant agreement no. 761445) and by the ests include performance optimization in wireless networks, mobile
Madrid Regional Government through the TIGRE5- traffic analytics, security and privacy, prototyping, and testbeds.
CM program (S2013/ICE-2919). Carlos Donato received his B.Sc in telecommunications engi-
neering (2013) and M.Sc in telematics engineering (2015) from
References UC3M. Currently, he holds a research position at IDLab — imec,
[1] 3GPP, “TS 23.501 — System Architecture for the 5G System; Antwerp, Belgium. His research interests lie in wireless com-
Stage 2,” Dec. 2017. munications (LTE and IEEE 802.11), energy efficiency, software
[2] P. Serrano et al., “Experimenting with Commodity 802.11 defined networking, and Cloud-RAN.
Hardware: Overview and Future Directions,” IEEE Commun.
Surveys & Tutorials, vol. 17, no. 2, 2015, pp. 671–99. Pablo Serrano [M’09, SM’16] received his degree in telecommu-
[3] M. Duarte et al., “Design and Characterization of a Full-Du- nication engineering and his Ph.D. from UC3M in 2002 and 2006,
plex Multiantenna System for WiFi Networks,” IEEE Trans. respectively. He has been with the Telematics Department of UC3M
Vehic. Tech., vol. 63, no. 3, Mar. 2014, pp. 1160–77. since 2002, where he currently holds the position of associate profes-
[4] P. D. Sutton, K. E. Nolan, and L. E. Doyle, “Cyclostationary sor. He has over 80 scientific papers in peer-reviewed international
Signatures in Practical Cognitive Radio Applications,” IEEE journal and conferences. He has served as a Guest Editor for Com-
JSAC, vol. 26, no. 1, Jan. 2008, pp. 13–24. puter Networks and on the TPC of a number of conferences and
[5] P. Bahl et al., “White Space Networking with Wi-Fi Like Con- workshops including IEEE INFOCOM and IEEE WoWMoM.
nectivity,” Proc. ACM SIGCOMM, 2009, pp. 27–38.
[6] N. Nikaein et al., “OpenAirInterface: A Flexible Platform for Yan Grunenberger received his Engineer and Ph.D. degrees
5G Research,” SIGCOMM Comp. Commun. Rev., vol. 44, from Grenoble University, France. He worked on iPhone mobile
no. 5, Oct. 2014, pp. 33–38. apps before doing research activities in various research institu-
[7] I. Gomez-Miguelez et al., “SRSLTE: An Open-Source Plat- tions (CTTC, 2010, and Telefonica Research, 2011, in Barcelo-
form for LTE Evolution and Experimentation,” Proc. ACM na, Spain). After doing product security at Qualcomm (2014),
WiNTECH, 2016. he again joined Telefonica (2015) to help create and run Tele-
[8] C. Capretti et al., “LTE/Wi-Fi Coexistence under Scrutiny: An fonica’s own end-to-end customer-centric lab where he is testing
Empirical Study,” Proc. ACM WiNTECH, Oct. 2016. and building PoCs around edge computing.