Home > Research > Scientific infrastructure

Scientific infrastructure

In order to support cutting-edge research, IMDEA Networks invests in the latest, state-of-the-art laboratories and laboratory test equipment, endowing the Institute with the capacity of transforming research into high added value products and services.

These laboratories are used for:

  • Constructing prototypes and measuring the devices, protocols and algorithms developed by the researchers.
  • Simulating complex base-band and medium access systems, as well as sophisticated radio subsystems.
  • Measuring radio parameters involved in mobile, fixed and satellite communications, designing and characterizing radiating elements, and measuring the effects on the radio electric spectrum of new protocols and algorithms designed by the Institute.

Examples of the laboratories capabilities include:

Global Computing Group

3D Printer/ Impresora 3D

This is a BQ Witbox2 3d printer, with maximum print speed up to 200 mm/s and a layer resolution up to 20 microns. This printer is used as support material for the different projects within the group.

Virtual Reality (VR) Kit / Kit de Realidad Virtual

The Global Computing Group has a Virtual Reality Kit consist of a Virtual Reality machine and HTC Vive Virtual Reality headset. The Virtual Reality Machine is a NVIDIA GeForce GTX 1070 based workstation with a Intel i5-6400K CPU 2.7Ghz x4 64bits, 40Gb RAM Memory DDR4-2133Mhz and a hard disk with 2Tb. The HCT Vive Virtual Reality headset including two 360-degree controllers, headset tracking, directional audio and HD haptic feedback. Moreover, the GCG lab has a reserved area used as virtual reality room.

Deep Learning Environment/ Entorno Deep Learning

The deep learning environment consists of two GPU computers and a NVIDIA Jetson TX2 board. The above image shows a 2x NVIDIA GeForce GTX 1080 Ti 11Gb GDDR5x based workstation, with an Intel i5-8600K CPU 3.6Ghz x6 64bits, 64Gb RAM Memory DDR4-2666Mhz, a SDD 250Gb and a HDD 1Tb..

This image shows a GeForce RTX 20880 Ti 11Gb GDDR6x based workstation, with a Intel I7-8700K CPU 3.7Ghz, 64Gb RAM Memory DDR4-3200Mhz, a SDD 250Gb and a HDD 8Tb. This workstation is shared with the Ubiquitous Wireless Networking Group.

Finally, the above image shows the NVIDIA Jetson TX2 development kit board that features an integrated 256-core NVIDIA Pascal GPU, a hex-core ARMv8 64-bit CPU complex, and 8GB of LPDDR4 memory with a 128-bit interface. The Jetson CPU combines a dual-core NVIDIA Denver 2 alongside a quad-core ARM Cortex-A57. All machines included in the Deep Learning environment are currently being used to training a neural network in order to obtain a fish classifier under the Symbiosis project.

Internet Analytics Lab

Golem testbed

Golem is the main computing resource used by the IAG (Internet Analytics Group - http://iag.networks.imdea.org/) at IMDEA Networks to run network measurement (both active and passive ones) and perform both dynamic and static analysis of Android software.

Opportunistic Architectures Lab

Monroe testbed

MONROE node consisting in two paired APU2 motherboards with three LTE cat4 modems and one WiFi adapter" that are running under Orange, Yoigo, Pepephone LTE coverage in Spain.

MONROE node: Installation box view.
MONROE Testbed at IMDEA Networks: Consists of 20 static nodes. 4 are development nodes and 12 testing nodes. These nodes are also connected through Ethernet alongside WIFI and LTE connectivity.

Dynamic Provisioning testbed

The provisioning tesbed consists of a DEL T640 Server with 40 CPU cores, 2.8GHz clock speed, 128GB RAM. It hosts virtual machines provisioned automatically using custom algorithms based on Machine learning. These Virtual machines serve external requests to the server and their number is dynamically adjusted depending on the incoming demand. The server therefore acts as a micro-cloud data centre receiving requests from cloud application providers represented here by the laptops running custom shell and python scripts. These are all interconnected in a Gigabit Ethernet LAN via a Cisco SLM2008T switch.

Pervasive Wireless Systems Group

Ettus B210 software-defined radio connected to eight port splitter
Hardware of outdoor spectrum sensor

 

NVIDIA GeForce RTX 2080 for neural networks
OpenVLC testbed with OpenBuilds ACRO moving structure

 

OpenVLC1.3 board
Spectrum sensor for accurate outdoor measurement

 

Testing ourOpenVLC board with the Agilent 3000 X KeySight oscilloscope

Ubiquitous Wireless Networks Lab

Underwater communications testbed

This is a collection of 4 underwater acoustic transceivers that operate in the 18-36 kHz band. They are connected to semi-rugged laptop computers through a switch and waterproof underwater cables. They can be operated both from a bench power supply (like in this picture) and from batteries.

Neptune, the horse power server of the underwater laboratory: This is a DELL PowerEdge server that is employed to run computer simulation for underwater communications, localization, and networking. It is also used as a remote control center for the UWN group's underwater acoustic transceivers.

Wireless Networking Group

Millimeter-wave SDR-based Open Experimentation Platform

Goals:

  • FPGA-based baseband processing system
  • IEEE 802.ad compliant transceiver
  • Higher than 2 GHz RF bandwidth
  • Flexible modular design
  • Digital and Hybrid beamforming using mm-Wave phased-array antennas
  • Localization and radar applications

Vadatech processing system

  • AMC599 board:
    • Xilinx KintexUltrascaleFPGA (>1.5 Million Logic Cells)
    • 20 GB of DDR4 memory space (3 memory banks)
    • Dual DAC @ 5GSPS (16 bits) and Dual ADC @ 6 GSPS (12 bits)
    • PCIe and 10GbE interfaces
  • AMC726 board:
    • Corei7 processor system
    • Physical connection to the AMC599 board through PCIe

Sivers60GHz phased-array system:

  • 16+16 Tx/Rx antenna array
  • Full 802.11ad channel bandwidth (2.16GHz)
  • 57-71 GHz coverage
  • li>Integrated (changeable) codebook
  • ~6º phase resolution
  • Controller through SPI interface.

mmWave band radio link

The Sivers IMA form the building block for mmWave band radio link applications. Each unit consists of one up- and one down-converter. The up- and down-converters operate independently. The boards are used as a reference design for e.g. an RF front end for a point-to-point or point-to-multipoint radio link. The evaluation board offers all necessity for interfacing the converter modules with the I/Q baseband inputs and outputs of a modem. RF output/input is over WR15 waveguide. The antennas provided by the lab are omnidirectional and directional with apertures of 7, 20, and 80 degrees.