As a world-leader in the design and supply of highly sophisticated digital signal processing boards and software-defined radio (SDR) solutions, Nutaq partners with Universities and research centers in the design and development of next-generation wireless communication systems. Nutaq’s development platforms, based on large FPGAs, are complemented with intellectual property (IP) functional blocks for use by users as starting points in their own designs.

Recently, there has been heightened interest for “massive” multiple-input/multiple-output (MIMO) development solutions – that is, platforms that support large antenna arrays (typically more than fifty antenna elements) along with the associated signal processing capabilities. New Massive MIMO technologies, targeted for the next 5G wireless infrastructure, promise a ten-fold improvement in spectral efficiency while achieving a hundred-fold gain in power efficiency. Currently, no prototype exists that can truly be termed “massive.” Some quasi-massive prototypes have been built (such as the 64-antenna Argos

[1] and 48-antenna Argos v2), but they have practical limitations and they do not support the data throughput required for real-time massive MIMO experiments.

Nutaq has partnered with the Université de Sherbrooke, in Quebec, Canada, to push the barrier further and develop a reference design based on their platforms.  This reference design will serve as a starting point for exploring more advanced large array transmit and receive architectures, by providing algorithms that can easily scale with current technologies and associated costs. The project will be lead by Prof. Sébastien Roy, which has considerable experience in communication theory, wireless communications, array processing, and associated implementation issues. He has designed and supervised the design of various array processing schemes and complete physical layer implementations within FPGA-based platforms. He has also supervised the development of custom advanced RF front-ends, including passive beamformers and RF-level combining. He holds several patents related to array processing and one related to synchronization and channel estimation in OFDM systems. He has recently published several papers related to simplified practical diversity combining schemes. The current project aims to implement these novel methods as well as extensions thereof and demonstrate their feasibility.

Using a 2×2 antenna OFDM receiver reference design as a starting point, the project consists in developing a 16-antenna receiver on Nutaq’s Massive MIMO Subset System.

To limit the complexity and leave some FPGA real estate for other functions (and for additional antennas eventually), it is critical to choose or design array processing algorithms that are simple and scale well with the size of the array. While 16 antennas is not enough to observe the massive effect, it is a starting point to explore larger-scale arrays and validate algorithm selection that can be easily scaled to system sizes on the order of 100×100.

The project will be divided into five stages. The results from each stage will be discussed in subsequent blog posts.

Stage 1: Developing expertise with the hardware platform and the OFDM reference design

Before starting on the design and implementation stages, the Sherbrooke researchers will spend a month working closely with Nutaq’s engineers to learn as much as they can about the M.MIMO Sub-System and the OFDM reference design.

Stage 2: Implementing the real-time channel emulator

Work will begin by first implementing the real-time channel emulator and transmitter logic onto the transmit sub-system, which provides enough D/A channels for the purpose of this stage.

Stage 3: Designing the receiver

The design work will commence on the receiver. The OFDM design will be extended to four antennas and, while assuming perfect channel knowledge and synchronization, will be augmented with various simple array processing schemes: selection diversity, multiple regression/correlation (MRC), minimum mean squared error (MMSE), and layered-space time (LST) MIMO. These can then be validated with the channel emulator.

Stage 4: Extending to 16 antennas

The core of the project will follow, namely the extension to 16 antennas and the implementation of additional schemes based on dividing the array into groups (such as four groups of four). The schemes may include, for example, hierarchical selection/MRC, hierarchical selection/MMSE, MMSE per block, and LST-MIMO with per-block processing.

Stage 5: Implementing channel estimation and synchronization algorithms

The final stage, if time allows, is to add appropriate channel estimation and synchronization algorithms with a view towards interfacing with additional platforms for building larger arrays.

What is Next

Nutaq is looking to use the results of this exploratory project as a starting point for launching multiple other initiatives in Massive MIMO technology, by combining its expertise with universities and research centers. Nutaq is determined to be a leading provider of Massive MIMO solutions.

References

C. Shepard, H. Yu, N. Anand, L. E. Li, T. L. Marzetta, R. Yang, and L. Zhong, “Argos: practical many-antenna base stations,” in Proc. ACM Int. Conf. Mobile Computing and Networking (Mobi- Com), Aug. 2012. [?] C. Shepard, H. Yu, and L. Zhong. “ArgosV2: a flexible many-antenna research platform.” Proc. Mobicom: 19th annual internat. conf. on Mobile computing & networking, Miami, Oct. 2013.