Real-time processing used by German research team in experimental wireless communication base station.
As 5G, the fifth generation of wireless communication standards, is currently being rolled out, scientists keep researching ways to improve the bandwidth efficiency of wireless communication networks. With the surge of connected devices (IoT) such as autonomous vehicles, future communication networks must be able to exchange data in real time with more and more terminals within a cell.
One of the promising technologies to achieve that is Massive MIMO base stations (Multiple Input, Multiple Output). Massive MIMO base stations consist in an array of tens or hundreds of antennas that communicate simultaneously with a large number of terminals within the cell. The signal received from each terminal in the cell has specific characteristics due to its relative position to the base station. SDMA (Space Division Multiple Access) is a method that analyzes these characteristics to optimize the downlink transmission accordingly. This ensures that the capacity of all antennas of the array is used at its maximum, delivering the best possible bandwidth to all terminals in the cell.
The key challenge in Massive MIMO antenna development is the processing of all these signals in real time. Since all terminals within the cell are connected to all antennas of the base station, a complex encoding and decoding must take place to put together the pieces of the puzzle of each uplink and downlink signal. And this processing must be as fast as possible, since mobile terminals change position permanently within the cell.
Pushing the limits of wireless communication
A research team of the department of Electrical Engineering and Computer Sciences (EECS) of the Technical University (TU) of Berlin, Germany led by Prof. Giuseppe Caire is working on solving these challenges. Prof. Caire is a world-wide recognized and distinguished scientist in communications research. The purpose of the research is to increase the capacity, functionality and reliability of wireless devices and systems by investigating new concepts and optimize them to determine their fundamental limitations. The team has built experimental massive MIMO antennas to prove the validity of their concept. The results of this research and its proofs of concept will be used by the industry to develop the mobile communication systems of the fifth and sixth generation.
Looking for a solution to process the incoming signals, the team at TU Berlin turned to Israel-based FPGA-processing expert Gidel. The experimental massive MIMO base station is able to communicate with eight user terminals, simultaneously, with less than the radio resources a conventional base station needs to communicate with just a single user terminal.
All antenna signals and communication streams are processed on a single high-performance FPGA board which is connected to the radio front-ends as well as to the network infrastructure via multi-gigabit transceivers and the PCI express interface. This solution ensures a high throughput and extremely low latency, which is a critical point in wireless communication. To achieve this, Gidel implemented Intel’s Arria10 FPGA because of their high floating point processing power and data communication performance.
Massive MIMO base stations with SDMA use beam forming to target the radio-emissions to each terminal. This reduces the radiation power needed for the transmission and hence the risk of interferences. Adding antennas to the array improves the precision of beam forming and hence the efficiency of the base station.
The experimental setup of TU Berlin, which was developed by Dr.-Ing. Andreas Benzin, features an array of 64 antennas for just one Gidel FPGA board for signal processing. In a practical implementation, the Gidel board could handle up to 192 antennas at once. Larger base stations could easily be implemented by adding FPGA boards to the system.
Easy upgrades with consistent hardware and software
Gidel has been at the side of the research team for many years. “It all started in 2005”, remembers Dr. Ing. Andreas Kortke. “At that time, Gidel’s ProcStar II boards were offering a large number of high speed general-purpose FPGA-signals and the ProcWizzard programming tool made it easy to program”. The team could focus on implementing their signal processing algorithms on the FPGA from the beginning. They didn’t need to waste time developing the peripheral interface to the host, drivers, etc., since these were provided by Gidel.
Over the years, the team built new experimental systems with newer generation FPGA hardware. “Migrating to the latest Gidel platform was extremely easy because the hardware concept remained the same and the same API suite was maintained from one platform to the other”, says Kortke.
The collaboration between Gidel and the research team at TU Berlin has been going on for more than 15 years. The technical capabilities of Gidel’s FPGA boards are not the only reason for this success, though. “The quality of the technical documentation of both hardware and software was also very helpful for our work”, says Andreas Kortke. “And we could always count on Gidel’s customer support in case of issue”. With technologies like Internet of Things (IoT), edge computing, autonomous vehicles, etc., the need for fast and reliable wireless internet will keep growing in the years to come. With high-speed processing and low latency, Gidel’s FPGA technology contributes to creating the infrastructures of tomorrow’s connected world. Reuven Weintraub, CEO of Gidel said: “This application demonstrates the potential of our FPGA technology whenever high throughput and real-time processing is required”.