In this series:

This blog post discusses massive MIMO from an implementation and prototyping perspective. A large number of new theoretical techniques and concepts have recently been proposed for many-antenna base stations. It is also being argued that a network capacity increase by a factor of a thousand is achievable when massive MIMO is used alongside dense small cell deployments

[1, 2]. Energy efficiency is also significantly improved, which makes for greener systems [3]. However, without experimental validation, it is difficult, if not impossible, to predict the practicality and performance of these proposed techniques on real hardware in complex, rapidly varying, real-world conditions [4]. In fact, there is a large demand for a flexible many-antenna development platform that supports rapid prototyping and validation of new massive MIMO concepts and techniques.

Hajim Suzuki from the CSIRO ICT Centre proposed a novel Ngara Access system in which the central access point (AP) is equipped with a uniform circular array (UCA) installed on a high tower while each user terminal (UT) is equipped with a directional antenna, free of clutter, providing a predominantly line-of-sight (LoS) channel environment. He showed through simulations that the spectral efficiency of the proposed system can be improved linearly as a function of the number of antenna elements at the AP, without increasing the total transmitting power, provided half-wavelength antenna spacing is maintained and user groups of four or more are used to avoid an ill-conditioned channel. Hardware demonstrations based on the proposed system have achieved a spectral efficiency of 20 bits/s/Hz in a rural environment and 67 bits/s/Hz in a laboratory environment at a lower UHF band [5].

Cerato has addressed the hardware implementation of a low-complexity, high-performance detector for  32×32 MIMO. The design can reach very high data rate, up to more than 170 Mbit/s with QAM64 with a bit error rate (BER) of 10-1.5 to 10-2. This constitutes a cost-effective improvement over basic detection schemes [6].

Argos is a comprehensive real-time implementation of a flexible base station architecture that is scalable up to thousands of antennas and able to serve tens of terminals simultaneously through multiuser beamforming. Using commercial off-the-shelf (COTS) radio modules like the WARP platform, Shepard implemented a prototype with 64 antennas that is capable of serving 15 terminals through zero-forcing and conjugate multiuser beamforming. Extensive experimental characterization using this prototype shows that the spectral capacity increases from 12.7 bps/Hz when using a single antenna to 85 bps/Hz when employing zero-forcing multiuser beamforming, and up to 38bps/Hz when employing the less computationally intensive conjugate multiuser beamforming, all while using a mere 1/64th of the single antenna’s transmission power [7]. The authors in [7] have showed that the spectral capacity grows nearly linearly with the number of base station antennas and the number of simultaneously served terminals, as suggested by theory [7].

These findings need further investigation. Ngo and Larsson showed in [8] that, with channel state information (CSI) estimated from uplink pilots, they can only reduce the uplink transmit power per user inversely proportionally to the square-root of the number of base station antennas. This is due to the fact that when they reduce the transmit power of each user, channel estimation errors become significant. They termed this effect “noise contamination”.

The bottom line is that most massive MIMO research has been restricted to theory based on analytical or simulation results. The lack of an experimental platform has prohibited the validation of theoretical results and understanding of the impact of real-world factors on performance.

Nutaq’s Kermode XV6 AdvancedTCA computer blade [9] can be used to build an 8×8 MIMO node as it has four mezzanine sites onto which standard VITA-57 high-pin-count FPGA mezzanine cards (FMC) like the Radio420 can be connected [10]. The Kermode XV6 packs eight Xilinx Virtex-6 SX475T FPGAs, which deliver an outstanding 8.8 TeraMACs solely from their DSP48E1 dedicated multiply-accumulate engines. Each FPGA interfaces with two DDR-3 SDRAM SODIMM modules capable of supporting up to 4 GBytes, for an aggregate memory capacity of 64 GBytes. This memory may be used to store intermediate results of memory-intensive algorithms like minimum mean square error, zero-forcing, and eigenvalue decomposition for channel estimation and signal detection, or for beamforming within the coherence time. The blade connects to Zone 2 (backplane) and Zone 3 (custom backplane or rear transition module) at rates exceeding 500 Gbps, lending itself for highly efficient clusters, with up to 128 FPGAs in a single chassis. This connectivity enables the addition of extra 8×8 MIMO blades to scale up to a 128×128 MIMO system or even more within an AdvancedTCA framework [9]. The Radio420 FMC, on the other hand, operates up to 3.8 GHz and enables access to a massive array of patch antennas mounted in a small area.

References

[1] M. Ahmed Ouameur, “Massive MIMO – Part 4: Massive MIMO and small cells, the next generation network,” http://www.nutaq.com/blog/massive-mimo-%E2%80%93-part-4-massive-mimo-and-small-cells-next-generation-network[2] Qualcomm, The 1000x Data Challenge, http://www.qualcomm.com/solutions/wireless-networks/technologies/1000x-data[3] http://www.greentouch.org/index.php?page=home[4] M. Ahmed Ouameur, “Massive MIMO – Part 3: Capacity, coherence time and hardware capability,” http://www.nutaq.com/blog/massive-mimo-%E2%80%93-part-3-capacity-coherence-time-and-hardware-capability[5] H. Suzuki, R. Kendall, K. Anderson, A. Grancea, D. Humphrey, J. Pathikulangara, K. Bengston, J. Matthews, and C. Russell, Highly Spectrally Efficient Ngara Rural Wireless Broadband Access Demonstrator,  in Proc. IEEE International Symposium on Communications and Information Technologies (ISCIT), October 2012.[6] B. Cerato and E. Viterbo, Hardware implementation of low-complexity detector for large MIMO, in Proc. IEEE ISCAS’2009, pp. 593-596, Taipei, May 2009.[7] 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 (MobiCom), Aug. 2012[8] H. Q. Ngo, E. G. Larsson, and T. L. Marzetta, “Uplink power efficiency of multiuser MIMO with very large antenna arrays,” in Proc. 49th Allerton Conference on Communication, Control, and Computing , 2011.[9] Nutaq inc., “KERMODE: AdvancedTCA Virtex-6 FPGA Compute Blade,” http://www.nutaq.com/products/kermode-xv6[10]  Nutaq inc., “Radio420X: SISO/MIMO FMC RF Transceiver Module,” http://www.nutaq.com/products/function/rf