OpenAirInterface Massive MIMO Testbed : A 5G Innovation Platform

OpenAirInterface Massive MIMO Testbed :  A 5G Innovation Platform

Xiwen Jiang, Florian Kaltenberger, Raymond Knopp, Houssem Maatallah1467471996


Massive MIMO is one of the key enabling technologies for next generation of wireless communication systems [1]. It has a strong potential in increasing network capacity, offering high spectrum efficiency, saving transceivers’ energy and many other advantages. Recently, Massive MIMO has attracted great attention from both the research and industrial community. The OpenAirInterface Massive MIMO testbed is built using the open source 5G platform. It is one of the world’s’ first LTE compliant base station equipped with large antenna array, which can directly provide services to commercial user equipment (UEs). It shows the feasibility of using Massive MIMO already in current generation LTE standard by using Transmission mode 7-10, indicating the possibility of smoothly evolving the wireless network from 4G to 5G. We show here an innovation platform for solving 5G Massive MIMO challenges, by giving the possibility of advanced algorithm testing, concept validation, channel measurements, etc.

Keywords: 5G, Massive MIMO, OAI, LTE, Open source

I. Introduction

The emergence of various new wireless applications and the growing number of connected devices are greatly challenging the current wireless network, not only in network capacity, but also in coverage and latency. The telecommunications industry is thus working towards 5G, the fifth generation of wireless network, to meet the large scope of challenges posed by future networks. Massive MIMO is one of the key 5G technology candidates, involving hundreds of antennas installed at the base station. It promises significant gains in wireless network capacity, spectrum efficiency as well as offering the possibility of greatly reducing energy consumption, enhancing the reliability and reducing latency [2]. The great potential of Massive MIMO has motivated many researchers and engineers to bring the technology from theoretical domain to practical deployments.

In order to validate theoretical concept and address the challenges and potential of Massive MIMO, prototyping is of essential importance. Hence, based on OpenAirInterface, we have built a LTE compliant Massive MIMO testbed. It integrates Massive MIMO technology into the LTE standard and thus can interoperate with commercial UEs using Transmission modes 7-10. To overcome the challenges of accurate channel state information (CSI) acquisition, the system is based on the channel reciprocity in a TDD system [3]-[5]. With its 64 antenna array, the system can serve up to 4 UEs on the same frequency-time domain resource.

Our testbed is one the world’s’ first real-time Massive MIMO testbed built using open source software, with the 3GPP LTE protocols implemented from the physical layer to the network layer. It demonstrates that Massive MIMO can be perfectly used in LTE and its usage in 5G can be a smooth evolution starting from the current 4G standard. It provides a platform based on which engineers can innovate and test their concepts on the challenges in making Massive MIMO more efficient, such as common channel beamforming, new reference signal design and the design of CSI feedback schemes.

Figure 1 : OpenAirInterface  Massive MIMO testbed: One of the world’s first real-time full LTE protocol stack compatible testbed

Fig. 1 : OpenAirInterface Massive MIMO testbed: One of the world’s first real-time 3GPP LTE compatible testbed

II. System Architecture

A. Synopsis

Figure 1 illustrates the flexible and scalable TDD based OpenAirInterface Massive MIMO system. It can support a large antenna array of 64 elements with 5MHz bandwidth at the frequency of 2.6GHz, serving up to 4 users on the same time and frequency resource. The bandwidth limitation can be easily removed by upgrading the FPGA of EXMIMO2 RF platform or by using other RF platforms such as Ettus USRP B210, Ettus USRP x310 [13] or LimeSDR [12]. OpenAirInterface natively supports multiple RF platforms (USRP B210, USRP x310, LimeSDR) which have support much higher transmission bandwidth and can be easily synchronized as shown below to support higher bandwidth Massive MIMO platform using the same software architecture. The key parameters of the system are summarized in Table 1.




Number of antennas
Up to 64
Center Frequency
Sampling Rate
FFT Size
Number of used subcarriers
Slot time
Maximum simultaneously served users
Currently 4 (LTE release 10), extendable
Table 1 : Key parameters of OpenAirInterface  Massive MIMO testbed

Figure 2 shows the system architecture, comprising a 64 antenna array, 16 ExpressMIMO2 Radio Frequency (RF) cards, a PCIe backplane, one clock distribution module and a high-end Xeon server running Linux and OpenAirInterface. In the following sections, we briefly describe the hardware and software components used in the testbed.

Fig. 2 : OpenAirInterface  Massive MIMO architecture

Fig. 2 : OpenAirInterface Massive MIMO testbed architecture

B. Hardware Components

1. ExpressMIMO2 card

ExpressMIMO2 card is a low cost hardware target enabling experimentation with OpenAirInterface [6]. It is developed by OpenAir5GLab@EURECOM and can be used by OpenAirInterface soft-modem to drive up to 4 parallel RF chains with up to 20 MHz bandwidth in the range of 350-3800 MHz.  It interconnects with a baseband computing engine using Gen1 1-way PCIe (2.5 Gbit/s peak full-duplex bi-directional throughput). The board is built around a low-cost Spartan-6 FPGA (150LXT) with native PCIexpress on the FPGA fabric and coupled with 4 high-performance LTE RF ASICs, manufactured by Lime Micro Systems (LMS6002D).  The combination allows for four full-duplex or half-duplex radios to be interfaced with a desktop or laptop PC without the need for external RF.

Fig. 3 : ExpressMIMO2 Cards

Fig. 3 : ExpressMIMO2 Cards

2. PCIe chassis

The testbed uses Magma’s ExpressBox 16 PCIe backplane for multiplexing 16 Gen1 1-way PCIe (40 GBit/s peak) into a single 16-way Gen2 PCIe link providing a peak data rate at 80Gbit/s [7]. It is used to host 16 ExpressMIMO2 RF cards, and is in charge of the communication between those cards with the Intel Xeon server.


Fig. 4: Magma’s ExpressBox 16  PCIe backplane

3. High-end Xeon Server

We use a high-end 20-core Xeon server (10-core dual-processor 3 GHz) with AVX2 instructions running a Linux real-time OS with low-latency kernel and OpenAirInterface5G LTE baseband modem.

4. Ettus Research Octo-clock

We cascaded two Ettus Research Octo-clock [8] in our testbed, each providing both time and frequency synchronization for 8 ExpressMIMO2 cards. A 61,44MHz clock signal and a pulse per frame (every 10ms) signal are generated by the master card and serves as the external input. These signals are then amplified and distributed by Octo-clock to synchronize the other slave cards.


Fig. 5: Ettus Research Octo-clock

5. Huawei Antenna array

The antenna array, provided by our partner Huawei Technologies (Paris), is composed of 20 patch antennas with 4 elements each. The patches are mounted on a rack with 4 antenna rows, each with adjustable height and tilt, as well as the possibility of moving patches in each row, thus offering a great flexibility for antenna element arrangement.  The antennas have been optimized for the 3GPP band 38 (2.6GHz, TDD, 50MHz bandwidth).


Fig. 6: Huawei Antenna Array

C. OpenAirInterface Software Platform

The software component of the testbed is built on OpenAirInterface, an open-source standard-compliant implementation of a subset of Release 10 LTE for UE, eNB, MME, HSS, SGw and PGw on standard Linux-based computing equipment (Intel x86 PC/ARM architectures) [9]. OpenAirInterface comprises of two repositories; OpenAirInterface5G and OpenAirInterfaceCN, dedicated to the radio access network (RAN) and the core network (CN) implementation respectively. OpenAirInterface5G can be used for simulation/emulation, as well as real-time experimentation on off-the-shelf software defined radio cards, like the aforementioned ExpressMIMO2 card but also the popular USRP from National Instruments/Ettus, LimeSDR, BladeRF, and other RF platforms. It comprises of the fully compliant LTE protocol stack from the physical to the networking layer and can interoperate with commercial LTE terminals and can be interconnected with OpenAirinterfaceCN or closed-source EPC (Enhanced Packet Core) solutions from third-parties. The objective of this platform is to provide methods for protocol validation, performance evaluation and pre-deployment system tests.

III. LTE complaint Massive MIMO Testbed

Massive MIMO is viewed as an enabler of the next generation of wireless communications. But it can also be applied to the current generation LTE standard. In fact, 3GPP has defined the notion of “Transmission Modes” (TMs) for different usage of MIMO in LTE, which can be categorized as transmit diversity, spatial multiplexing and beam-forming. TM 7 is defined in Release 8, where an arbitrary number of physical antennas at base station can be used as a logical antenna port (port 5) to create a narrow beam towards the targeted user. Release 9 extends  TM 7 to TM 8, giving the possibility of transmitting a dual stream to a single or two users, whereas in release 10, this is further extended to TM 9 where up to 8 layers for a single user transmission and up to 4 layers for multi-user transmission is supported. Release 11 adds TM 10, similar to TM 9 with up to 8 layers transmission but the transmit antennas can physically locate on different base stations. In Release 13, no new transmission mode is defined, but CSI Reference signal (RS) has been extended to 16 ports [10]. Moreover, the ongoing work item in release 14 [11] on the enhancement of Full-Dimension MIMO (special case of Massive MIMO in 3GPP) for LTE has defined the objective of extending the CSI-RS to 32 ports with enhancement on CSI reports and support for providing higher robustness against CSI impairments.

The OpenAirInterface Massive MIMO testbed relies on the implementation of TM 7-10 to use the large number of antenna array. Thanks to the channel reciprocity property in a TDD system, it acquires accurate downlink channel state information directly from the estimation of uplink pilots, based on which it can perform advanced beamforming algorithms to different UEs. The testbed uses a cost effective “Over-The-Air (OTA)” calibration method [3]-[5] to compensate the hardware asymmetry which breaks down the channel reciprocity and can achieve near optimal beamforming performance.

Currently only TM 7 has been implemented. Commercial UEs supporting this transmission mode can be connected to the massive array base station, and can surf internet through our Massive MIMO testbed. We hope that with our community support, we can extend OpenAirInterface towards future 3GPP releases thus implementing other transmission modes such TM 8-10 and also integrate successfully with other supported RF platforms (USRP B210, USRP x310 and LimeSDR).

IV. Conclusions

OpenAirInterface Massive MIMO Testbed is one of the world’s first real-time full LTE protocol stack compatible Massive MIMO system. Having open source platform OpenAirInterface as its core, the testbed can interoperate with commercial UEs, showing the possibility of using Massive MIMO in current generation LTE standard and beyond. It opens up the possibility of testing future innovations in a real and practical environment and will certainly have a significant impact in helping researchers’ and engineers towards designing 5G cellular systems.


We would like to bring our special thanks to Huawei Technologies (Paris) for their technical and financial support.


[1] T. L. Marzetta, “Noncooperative cellular wireless with unlimited numbers of base station antennas,” IEEE Trans. Wireless Commun., vol. 9, no. 11, pp. 3590–3600, Nov. 2010.
[2] E. G. Larsson, O. Edfors, F. Tufvesson, and T. L. Marzetta, “Massive MIMO for next generation wireless systems,” IEEE Commun. Mag., vol. 52, no. 2, pp. 186–195, Feb. 2014.
[3] F. Kaltenberger, H. Jiang, M. Guillaud, and R. Knopp, “Relative channel reciprocity calibration in MIMO/tdd systems,” in Future Network and Mobile Summit, Florence, Italy, Jun. 2010.
[4] X. Jiang, M. Cirkic, F. Kaltenberger, G. L. Larsson, L. Deneire, and R. Knopp, “MIMO-TDD reciprocity and hardware imbalances: Experimental results,” in Proc. IEEE International Conference on Communications (ICC), London, United Kingdom, Jun. 2015.
[5] X. Jiang; F. Kaltenberger; L. Deneire, “How accurately should we calibrate a massive MIMO TDD system?” International Conference on Communications (ICC), Workshop, May 23-27, 2016, Kuala Lumpur, Malaysia
[6] OpenAir5GLab EXPRESSMIMO2 RF Platform,
[7] Magma PCIe Chassis,
[8] Ettus Octo-clock,
[9] OpenAirInterface Software Alliance,
[10] 3GPP Technical Specification 36.211 “Evolved Universal Terrestrial Radio Access (E-UTRA); Physical Channels and Modulation (Release 13)”,
[11] RP-160623, “Enhancements on Full-Dimension (FD) MIMO for LTE”, RAN1 #71, Göteborg, Sweden.
[12] LimeSDR,
[13] Ettus B210, x310.