An Advanced Hardware Platform

to verify 5G Wireless Communication Concepts

Thomas Wirth, Matthias Mehlhose, Jens Pilz, Ralf Lindstedt
Dennis Wieruch, Bernd Holfeld and Thomas Haustein
Fraunhofer Institute for Telecommunications, Heinrich Hertz Institute, Einsteinufer 37, 10587 Berlin, Germany
Email: thomas.wirth@hhi.fraunhofer.de

Abstract—Verification of first 5G concepts of a new air inter- Furthermore, device-to-device communication as in smart grid
face with real hardware prototypes is challenging. In this paper, and the Industrial Internet will drive reliability and latency
we highlight a flexible software-defined radio (SDR) platform, requirements. Fast machine control loops and wireless emer-
which can be used to implement novel 5G communication
concepts. The proposed software and hardware architecture was gency stop functionality requires delays below 1 ms. With the
previously used to implement early LTE-Advanced features, e.g. number of sensors increasing, the energy costs and efficiency
flexible multi-band carrier-aggregation (CA) and studies on new has to be revolutionized, such that a 10x longer battery life
multi-carrier waveforms. The proposed hardware setup allows can be achieved and the energy cost per bit is reduced by a
early proof-of-concepts (PoC) by extending hardware-in-the- factor of 100 to a 1000-fold.
loop (HIL) experiments to small laboratory experiments and
finally real outdoor test trials of a novel 5G air interface and B. Key Concepts for 5G Prototyping
concepts.
Index Terms—5G Prototyping, Remote Radio Unit, Novel Air 5G systems and subsystems will require the development
Interface, Software-defined Radio Platform of a total new chip generation for signal processing of analog
and digital signals. Higher integration of analog and digital
I. I NTRODUCTION circuits will make chips more power efficient with support for
Current understanding of 5G directed research targets pri- more frequency bands as well as processing capabilities for
marily the next generation air interface, core network and decoding. In addition, signal processing will handle increas-
the underlying system architecture. While novel concepts ingly more bandwidth and interfaces between layers will scale
and mechanisms are being developed, there is an increasing from 1 Gbps to 10 Gbps inside the base stations and multiple
demand to demonstrate and evaluate key mechanisms with 10 Gbps links which can realize 40-100 Gbps in total between
real hardware. First experiments and lab tests will identify the base station network comprising fronthaul and backhaul
challenges and highlight limitations. These lab tests will later over fiber and suitable wireless extensions.
be extended to field trials while the technology is becoming Furthermore cloud technology with scalable base band
more mature. Since new transmission concepts with high com- signal processing platforms using off-the-shelf computer hard-
plexity are resource demanding, implementation on powerful ware architectures will gradually take over more data and
software-defined radio (SDR) platforms is a key for efficient signal-processing tasks currently performed on dedicated and
early prototyping and benchmarking. optimized processors primarily supporting PHY and MAC
In this paper, we will highlight the key challenges for pro- functionalities of the wireless communication layers.
totyping of the next generation mobile radio standard or 5G. In the next subsections, we will highlight a selection of key
We will propose a very flexible and powerful SDR hardware conceptual changes compared to current 4G settings, which
platform and illustrate exemplarily which forthcoming wireless have to be addressed in 5G.
technology can be implemented on the proposed platform. 1) Flexible Digital Signal Processing: Beyond content
II. 5G C HALLENGES AND C ONCEPTS distribution and proliferation in closer proximity of the users,
flexibility in moving functions and signal processing capa-
A. 5G Use Cases and Requirements bilities within the network requires substantial changes in
The requirements currently being discussed for 5G are a digital hardware architectures. Allowing more OSI stack layer
1000-times gain in capacity targeting very high data rates signal processing being performed at various sections of the
of up to 10 Gbps. With the deployment of the Internet-of- wireless network requires high-speed interfaces and protocol
Things (IoT), the number of devices communicating with each termination at new endpoints. This flexible signal processing
other will dramatically increase over the next years. With can range from baseband signal processing pooling in C-
these high capacity demands, service delivery for dense user RAN fashion plus a powerful fronthaul network for distributed
crowds and dense sensor systems will define 5G requirements. remote radio units at the antenna locations, to fully distributed
in network processing types of wireless communication and
This work was partially funded by the Federal Ministry of Education and
Research (BMBF) of Germany in the framework of the Cognitive Mobile data processing supporting concepts like network coding and
Radio (CoMoRa) project under support grant 16BU1200. compressed sensing techniques in a fully fledged way.

978-1-4799-8088-8/15/$31.00 ©2015 IEEE

 2) Digital Hardware Interfaces: The above sketched vi- trum, e.g. 15/24/28/40/60/70/80/90 GHz and above including
sion for higher flexibility in placing and moving signal- licensed and unlicensed bands. A unified approach could be
processing capabilities will allow more flexibility splitting realized using the LSA framework [2].
PHY and MAC signal processing chains enabling flexible 6) Licensed Shared Access: ASA/LSA is a novel way of
Fronthaul/Backhaul/X-Haul data rate and latency matching, authorising spectrum currently without changing ownership, in
and the support of fundamental mechanisms like user- and addition to licensed and license-exempt (unlicensed), see e.g.
control-plane splitting allowing new types of data and signal EU RSPG: Report on collective use of spectrum (CUS) [3] and
path settings across heterogeneous network topologies and air- ECC Report 205 on licensed shared access (LSA) adopted in
interfaces. Feb 2014 [4]. The authorized shared access (ASA) concept as
3) Analog and mixed signal processing: Bringing digital initially introduced by NOKIA and Qualcomm in 2011 and
signal processing closer to the antennas provides a new level in the meantime referred to as licensed shared access (LSA)
of signal processing substituting established analog control in standardization and regulatory bodies. These mechanisms
loops and filters by digital representations in dedicated chips. have to be extended have to include more frequency bands
Such trends will allow to exploit novel and adapt estab- and band combination. Handset chipsets need enhancements
lished concepts like Full-Duplex communication or digital pre- to detect interference from incumbent system. In addition,
distortion (DPD) in order to achieve e.g. diplex free multi- the 5G protocol stack has to implement signaling channels
band transmission, simultaneous FDD and TDD operation of to dynamically adjust resource scheduling in the incumbent
adjacent bands, coexistence of true dual wireless connectivity zones.
of user terminals to the network or in direct mode with 7) Air-Interface and new Waveforms: Current discussions
other devices. These opportunities come along with new on potential new 5G waveform candidates [5] are ongoing
challenges when it comes to very high numbers of antennas taking into consideration that 5G will be beyond a next level of
to be jointly used for signal processing and/or when the link level data rates into the Gbps range. New waveforms have
number of simultaneously communicating devices per area to provide umbrella functionality for energy efficient sporadic
increases significantly and requirements towards synchronicity communication, high speed data downloads, very high number
are very different compared to current single air-interface fully of users per cell or area, multi-service operation using the
synchronized approaches. Shrinking the many spectrum bands, same or different wireless networks in adjacent or the same
band combinations and many antenna components in size and spectrum. Challenges range from flexibility of radio resource
cost is a manufacturing challenge to be addressed when link scalability, via robustness under coarse timing and frequency
and systems performance gains are better understood. synchronization, near far effects, neighbor channel crosstalk
4) Active Antennas MIMO: MIMO introduced in WiFI, to robustness against hardware impairments.
WiMAX and LTE is now moving from 8 antennas, via multiple
of 8 antennas, up to massive MIMO (MMIMO) setups with III. F LEXIBLE SDR H ARDWARE
hundreds of antennas. Since antenna panel size matters, the Our software-defined radio (SDR) hardware prototype ap-
number of active antenna elements will increase with higher proach follows classical functional splitting into two key
frequency making MMIMO technology an enabler for spatial components: a flexible RF-radio unit and a multi-core signal-
multiplexing in the classical IMT bands and a must for longer processing platform, see Fig. 1.
range wireless communication links in the millimeter wave
spectrum to allow antenna gains from beamforming. A higher
flexibility of spectrum usage, enabled by key concepts like
LSA, see [1], allows to access and use a wide range of
frequencies. Here, frequency resources are aggregated using
secondary component carriers. LSA user terminals can either
handover to the newly aggregated spectrum or can be served
on primary and secondary carriers simultaneously. Allocation
of the particular frequency resources and bands depends on Fig. 1: Key components of SDR prototype: DSP-based signal
which type of service has to be provided to the LSA user: Low processing platform for baseband signal processing and RF-
frequencies preferably selected for mobile and deep indoor radio unit connected via high-speed multi-gigabit interface.
signal distribution or for building up large umbrella cells,
higher frequencies for lower mobility and shorter range, and
mmWave frequency for wireless back-/fronthaul as well as A. RF Radio Unit
short range communication by pushing access capacity in The radio unit supports various high-speed interfaces.
mmW small cells into the multi-Gbps range. The implemented interfaces are Gigabit Ethernet (GbE),
5) Spectrum: Using frequency spectrum as efficient and optical common public radio interface (CPRI) running at
flexible as possible: Flexible spectrum usage in many bands, 2457.6 Mbit/s over several km of fiber, as well as 10 GbE.
below 6 GHz as well as above 6 GHz, e.g. in the sub-10 CPRI implements a high-speed bus to transmit 15-bit I/Q
GHz range, the microwave band up to millimeter wave spec- samples between remote radio unit (RRU) and baseband signal
processing hardware. 1 Byte per I/Q data block is used to No. Antennas 2 TRx or 4 TRx
transmit control and management (C&M) data. This allows (TRx = transceiver) extendable to 8 TRx
remote configuration of the RRU as well as reception of ADC / DAC 12-bit
measurement data such as automatic gain control (AGC) Carrier Frequency Tunable: 70 MHz..6 GHz
values and received signal strength indicators (RSSI) for the Analog Bandwidth < 200 KHz .. 56 MHz
configured frequency band. The CPRI C&M channel allows Noise Figure < 2.5 dB
to change RRU settings on a LTE radioframe basis, e.g. every Modulation QPSK, 16-QAM, 64-QAM, 256-QAM
10 ms. In this configuration, the RRU can be used as a Frequency Duplex Mode FDD and TDD
frequency scanning device. Finally, I/Q compression allows to Diplexing SAW or external cavity filters
reduce I/Q data by a factor of 2 [6]. This enables support for Power Consumption < 30 Watts
higher bandwidth, up to 40 MHz, or the use of more antennas, Form Factor MicroTCA (single-width, full-height)
e.g. 4x4 MIMO, using the same CPRI interface rates. GbE over MicroTCA backplane
The clock concept for CPRI allows clock recovery on FGPA Xilinx Spartan-6 or Xilinx Zynq
the radio unit itself. Therefore, the RRU does not require Output power -8 dBm, external power amplification
expensive oscillators itself. The clock signal is derived from with 20-30 dBm output power
a high precision clock generated on the baseband card. This depending on carrier frequency
allows remote configuration of several RRUs from the same TABLE I: Basic configuration parameter sets and versions of
baseband card. For collocated RRUs, clock in- and outputs the {2, 4, 8}-TRx RF-radio unit.
can be used to synchronize several RRUs using master-slave
modes. The most important configuration parameters of the
RRU can be found in Tab. I. The basic schematic of the 4x4
RRU with 2 CPRI connectors is shown in Fig. 2. GbE as well as CPRI interfaces directly to both DSPs. GbE
and CPRI can be used via MicroTCA backplane or via
SFP modules on the frontpanel. In addition, this SDR board
implements a very flexible clock concepts with external clock
in- and outputs, a GPS, as well as a voltage controlled oven
oscillator (VCXO).
Beyond the currently used DSP platform for baseband
signal-processing, the RF-radio unit can be directly connected
via 40 Gbps Ethernet to a cloud-RAN capably multi-processor
unit, e.g. Intel server blade enabling aggregation of signal-
processing power into one unit for centralized processing
across signals transmitted/received by multiple RF-radio units.
Fig. 2: Block diagram of 2x2/4x4 SDR-RF: Stacked digital
and analog cards. IV. I MPLEMENTATION OF 5G C ONCEPTS
The SDR platform can be used to implement novel radio
The latest version of the RF-radio unit is equipped with systems with flexible frequency usage such as spectrum shar-
8TRX and the latest Xilinx Zynq FPGA and complies to the ing concepts or cross-layer scheduling, see [7]. Furthermore, it
MicroTCA form factor. In addition, it allows aggregation and can be used for PoCs of distributed antenna systems or novel
synchronization of several 8-transceiver (8-TRx) MIMO cards 5G multi-MIMO systems cascading several 8-TRx radio units
to support e.g. massive MIMO (MMIMO) antenna arrays with towards massive MIMO (MMIMO). The SDR platform was
64 antenna elements. The frequency can be tuned between previously used to implement an licensed shared access (LSA)
70 MHz and 6 GHz in FDD or TDD mode or to support user terminal as well as for transmission of arbitrary wave-
32 antenna elements operated in Full-Duplex mode supporting forms. Next, we will introduce the LSA concept and describe
beamforming at the same time. Furthermore, due to the two SDR use cases.
processing capability of the high performance FPGA and suffi-
cient memory, flexible PHY and MAC splitting options can be
A. Flexible Spectrum Usage
implemented allowing optimal matching of backhaul/fronthaul
needs within the particular experimental setup. Carrier aggregation (CA) is the key feature for licensed
shared access (LSA). The SDR prototype was used to im-
B. Baseband Processing plement a CA mobile terminal (MT) which can aggregate up
The baseband card has two TIC6670 DSPs with high-speed to 4 continuous or non-continuous component carriers, each
antenna interfaces (AIFs). Each DSP has 4 C66x cores and with 2x2 MIMO and 20 MHz bandwidth. The peak throughput
2 GB of DDR3 memory. Furthermore, the DSP board has on IP-layer of this MT is 600 Mbps and the SDR prototype
a Spartan-6 FPGA which runs Linux to load firmware and is depicted in Fig. 3. The software runs the real-time LTE
configure a variety of interfaces. It supports PCIe, SRIO, physical layer protocol implemented on the TI DSPs.
 0
Power = -2.1 dBm OFDM Signal
FBMC Signal
-20

Power Spectral Density [dBm/kHz]

-40

ACLR (OFDM) = -43.5 dB ACLR (OFDM) = -43.3 dB

-60
ACLR (FBMC) = -61.2 dB ACLR (FBMC) = -60.3 dB

-80

Fig. 3: Full SDR LTE mobile terminal for carrier aggrega- -100

tion (CA) or multi-MIMO implementation.

2620 2630 2640 2650 2660 2670 2680
Frequency Spectrum [MHz]

Fig. 5: Different waveforms and their adjacent channel leakage

ratio (ACLR) at 2.65 GHz measured using the SDR platform.

C. Massive MIMO
Key concept behind MMIMO is exploitation of directivity
of spatial signatures of user links allowing crosswise in-
terference nulling, which was first described in [8]. Recent
simulation studies show the gains of MMIMO in system-level
environments, see [9] and [10]. Similar to MIMO technology
10 years ago, many technical challenges have to be tackled for
successful early prototyping and proof-of-concept including
a wide range of systemic issues. These aspects include tight
synchronization of analog front-ends to support stable spatial
nulling, integration of antennas elements together with power
Fig. 4: Measurement result for LTE-A carrier aggregation with
amplifiers, LNAs, diplex-filters, efficient channel estimation
3 component carriers, 20 MHz bandwidth, 2x2 MIMO for
for high number of antennas, TDD vs. FDD, full-duplex trans-
continuous and non-continuous carrier aggregation realizing
mission to avoid diplexers, efficient group antenna selection
sum rates of approx. 450 Mbps.
for hybrid beamforming, frequency dependency and hardware
impairments measurement and compensation. Finally, form
factor and cost efficiency determine the feasibility of proto-
B. New Waveforms typing complex MMIMO systems.
Fig. 6 shows a MMIMO antenna array example designed for
The 5G SDR prototyping platform allows transmission the 3.6 GHz band which was recently setup at Heinrich Hertz
and reception of arbitrary waveforms. Waveforms can either Institute. The antenna cube with 8x8 active antenna arrays on
be generated in real-time, or recorded and stored on the 3 sides is to be used for densely populated areas with moving
baseband platform or radio head itself. This mode allows user clouds like e.g. train stations, shopping malls or trade
continuous playback of waveforms in a loop mode, as well show scenarios.
as online recording of over-the-air transmitted radio signals.
The received I/Q samples can be fully or partially decoded on
the baseband board or directly recorded via encapsulated bit-
stream to an external PC. The recorded bit streams can then
be processed offline using standard signal processing libraries.
This SDR prototype can transmit any waveform which
have been generated in a pre-defined standard I/Q format. In
particular, it can transmit any multicarrier signal, e.g. OFDMA,
SCFDMA, or filter-bank multicarrier (FBMC) signals. This
can be used to study various waveform characteristics, e.g.
adjacent channel leakage ratio (ACLR) and pre-distortion
techniques, see Fig. 5. In addition, standard LTE control
signaling can be embedded into new waveforms, which allows
to study novel synchronization, equalization, and decoding Fig. 6: Massive MIMO (MMIMO) antenna cube with
algorithms. 10x10 antenna patches with 8 active antennas per column.
 Each 8x8 antenna patch array is currently being assembled
and uses 8 RF-radio units, one for each 8-antenna column.
After initial lab tests, the MMIMO unit will become an integral
part of the 5G Berlin testbed facilities proving an open 5G
research and over the air field trials facility [11].
V. C ONCLUSION
We gave an overview about use cases and requirements
of the forthcoming 5G wireless communication system. We
highlighted an advanced hardware platform which can be used
to prototype and verify novel 5G concepts. Furthermore, we
showed how the proposed prototyping platform based on a
software-defined radio (SDR) solution can be used to realize
advanced MMIMO systems with more than 32 or 64 antennas.
The key enabler to realize this is the advanced RF-unit, which
can be re-configured to a wide frequency range. Finally, the
proposed hardware platform has the required flexibility and
advanced processing power for 5G wireless prototyping.
VI. ACKNOWLEDGEMENT
The authors would like to thank Andreas Forck, Hol-
ger Gäbler, and Kirsten Krüger for their contribution to the
hardware and firmware design as well as implementation of
key features on the radio unit.
R EFERENCES
[1] T. Wirth, B. Holfeld, D. Wieruch, R. Halfmann, and K.-J. Friederichs,
“System level performance of cellular networks utilizing asa/lsa mech-
anisms,” in Cognitive Cellular Systems (CCS), 2014 1st International
Workshop on, Sept 2014, pp. 1–5.
[2] M. Mueck, T. Haustein, and K. Sakaguchi, “Global standards enabling
a 5th generation communications system architecture vision,” in IEEE
Globecom 2014, Austin, TX, USA, 2015.
[3] RSPG 11-392, “Report on Collective Use of Spectrum (CUS) and other
spectrum sharing approaches,” Radio Spectrum Policy Group, European
Commission, Tech. Rep., Nov. 2011.
[4] CEPT ECC Report 205, “Licensed Shared Access (LSA),” Feb. 2014.
[5] G. Wunder, P. Jung, M. Kasparick, T. Wild, F. Schaich, Y. Chen,
S. Brink, I. Gaspar, N. Michailow, A. Festag, L. Mendes, N. Cas-
siau, D. Ktenas, M. Dryjanski, S. Pietrzyk, B. Eged, P. Vago, and
F. Wiedmann, “5gnow: non-orthogonal, asynchronous waveforms for
future mobile applications,” Communications Magazine, IEEE, vol. 52,
no. 2, pp. 97–105, February 2014.
[6] S. Nanba and A. Agata, “A new iq data compression scheme for front-
haul link in centralized ran,” in Personal, Indoor and Mobile Radio
Communications (PIMRC Workshops), 2013 IEEE 24th International
Symposium on, Sept 2013, pp. 210–214.
[7] B. Holfeld, R. Mochaourab, and T. Wirth, “Stable Matching for Adaptive
Cross-Layer Scheduling in the LTE Downlink,” in IEEE Vehicular
Technology Conference (VTC Spring), June 2013, pp. 1–5.
[8] T. Haustein, C. von Helmolt, E. Jorswieck, V. Jungnickel, and V. Pohl,
“Performance of mimo systems with channel inversion,” in Vehicular
Technology Conference, 2002. VTC Spring 2002. IEEE 55th, vol. 1,
2002, pp. 35–39 vol.1.
[9] M. Kurras, L. Thiele, and T. Haustein, “Interference aware massive
sdma with a large uniform rectangular antenna array,” in Networks and
Communications (EuCNC), 2014 European Conference on, June 2014,
pp. 1–5.
[10] M. Kurras, L. Thiele, and G. Caire, “Interference mitigation and mul-
tiuser multiplexing with beam-steering antennas,” in ITG Workshop on
Smart Antennas (WSA), March 2015.
[11] http://www.5GBerlin.de, “An open testbed facility for 5G research for
testing access, photonics, core, SDN and M2M technologies.” Berlin,
Germany, 2015.