08 Aug 2014

MIMO Signal Processing Considerations

David Hawke, Director Wireless Product Marketing, Xilinx looks at the signal processing requirements and considerations behind MIMO and how they can be optimised.

In a bid to meet demand for greater bandwidth, operators are tackling network expansion through the use of Multiple-In-Multiple-Out, MIMO antennas, an increasing number of mobile stations and wider range of frequencies.

But this demand cannot be met without adhering to commercial pressures; in the face of mounting operating costs, operators are also desperate for equipment that offers greater efficiency and higher integration.

This burden falls to the equipment manufacturers, who must now deliver these gains within smaller budgets and market windows. These design considerations for MIMO signal processing are of great importance in the design of any system and should be determined at the earliest stages oft he design process.

Cellular applications

The LTE rollout (both TDD and FDD) across the globe has come with its own unique requirements in terms of equipment features and performance, particularly when compared to previous air-interface roll-outs.

The most common configuration used for the CDMA2000 and WCDMA 3G rollout was a 20MHz 2x2 radio, a configuration that has now evolved to deliver wider bandwidth and support more frequency bands. Manufacturers are now commonly developing MIMO radios capable of delivering as much as 100MHz of usable bandwidth. These new configurations typically employ 4x4 and 8x8 MIMO antenna arrays.

Development of these ultra-wide-band radios will allow network operators to deploy a single radio that supports multiple frequency bands of interest - for example 1800MHz, 1900MHz, and even 2100MHz in the future—resulting in fewer remote radios mounted on the tower top, which in turn reduces costs. This evolution not only saves significant capital equipment costs through the use of fewer radios, but also significantly reduces operating costs and site-rental costs through wind-loading reduction and reduced total weight loading on the tower.

MIMO processing requirements

A 100MHz 8x8 radio requires 20x more signal processing than a 20MHz 2x2 radio. When considering the digital and analogue processing requirements for such radio equipment, it is hard to see how the products could ever be realised while meeting mechanical, thermal, and cost constraints imposed by passively cooled remote radios. Clearly, continued evolution of new generations of digital and analogue ICs is required.

Consider the system-level requirements for implementing a 100MHz 8x8 radio as an example.

In an 8x8 mobile radio, the dimensions that define the required FPGA devices needed include the number of antennas, the air interface, the number of carriers, and the instantaneous bandwidth. The number of antennas defines the required number of DAC and ADC connections. The radio’s instantaneous bandwidth defines the sample rate and therefore defines the speed and number of SerDes ports required to connect the digital signal processing in the FPGA with the DACs and ADCs.

Data converters have advanced significantly over the last few years with many manufacturers now offering multiple DACs and ADCs within a single package, connected to the digital radio processing circuitry via JESD204B. The JESD204B interface standard replaces the older, previously used LVDS parallel interfaces with high speed SerDes ports capable of running at 12.5Gbps. JESD204B interfaces significantly reduce the number of required connections between the DACs and ADCs and the digital front end processing circuitry. The result is fewer PCB layers and reduced interface power. Furthermore, these JESD204B converters are now becoming even more integrated; they are absorbing entire uplink and downlink RF-processing subsystems, resulting in high-performance RF ICs that include the DACs, ADCs, filtering, and modulators needed to create highly integrated and high-performance analogue front ends, AFEs.

More variables

In addition to determining the number of required SerDes transceivers, a mobile radio’s bandwidth and antenna count also dictate the amount of logic and DSP required for processing algorithms, including Digital Up Conversion, DUC and Digital Down Conversion, DDC. DUC and DDC algorithms are very well suited to FPGA implementations because they allow network customers to tweak the algorithms as needed without hardware changes. Two additional and commonly used algorithms are Crest Factor Reduction, CFR, and Digital Pre-Distortion, DPD, which improve a radio output power amplifier’s efficiency.

Based on the DSP resources required for implementing wideband radios and the number of SERDES required to communicate with eight antennas, a cost-effective method of implementing a mobile radio using current-generation Xilinx FPGAs would be to split the algorithmic processing between two devices, as shown in Figure 1. This system partitioning effectively creates two 4x4 100MHz radios. A by-product of this partitioning is that it spreads the operating power evenly across two digital devices, easing the thermal constraints and simplifying the design’s mechanics.

MIMO processing example

Figure 1: 100MHz 8x8 Radio implementation example using two Kintex-7 FPGAs

However, this system-level partition is no longer needed using Xilinx Kintex UltraScale devices, which have significantly more DSP capability, a larger number of SERDES ports, and consume significantly less power.

UltraScale architecture for MIMO

In fact, the UltraScale architecture has been highly tuned to the needs of this type of radio design. The UltraScale architecture’s highly efficient block RAM and LUT (Look Up Table) RAM and its optimised DSP48E2 DSP blocks efficiently combine to create the many types of filters needed for DUC, DDC, CFR, and DPD signal processing. The UltraScale architecture’s programmable-logic fabric and on-chip memory can implement high-performance, soft microprocessors, which can be used with hardware acceleration to implement highly scalable, low-footprint coefficient engines for DPD processing.

UltraScale FPGAs can achieve clock rates in excess of 491MHz, which permits the exploitation of time-division multiplexing to further reduce the amount of on-chip area devoted to signal processing while retaining the low power consumption of an FPGA-based radio design. Consequently, a 100MHz 8x8 radio can be implemented with just one Kintex UltraScale FPGA, as shown in Figure 2.

MIMO processing example

Figure 2: Comparable 100MHz 8x8 Radio implemented with one Kintex UltraScale FPGA

To implement such a complex radio in a single device, the overall operating power must drop considerably to reduce the generated heat. FPGAs based on the Xilinx UltraScale architecture have been tailored for low-power operation in several ways. First, the static power of two 28nm FPGAs is replaced by just one 20nm FPGA. The one FPGA device shown in Figure 2 consumes less power than the two FPGAs shown in Figure 1 while implementing the same number of channels.

Design considerations for low power

Furthermore, careful design has significantly reduced the dynamic power of the devices’ programmable-logic fabric and SERDES transceivers. Fewer of DSP blocks are needed for digital filter implementations based on complex MAC (multiply/accumulate) operations. Reducing the required number of DSP blocks reduces the dynamic power and area needed to implement DUC, DDC, CFR, and DPD algorithms.

The system-level result is an 8% reduction in cost per transmit/receive pair and a power reduction of more than 31% relative to an implementation based on a previous-generation device. Additional system-level cost savings can be realised through the reduced number of required PCB layers and reduced power supply complexity. A smaller power supply and smaller, less complex cooling hardware further reduce the remote radio system’s weight and the required enclosure size, which further reduces BOM and operating costs.

The demand for radio infrastructure equipment that matches lower purchase and operating costs with higher reliability is rising; the key to meeting that demand is integration. As shown here, ultra-wideband MIMO radios offer the solution but, until recently, could only be implemented using multiple discrete components which threatened to add cost and power. Through the higher integration offered in modern FPGAs, the MIMO landscape is changing.

Page 1 of 1

About the author

David Hawke is the Director for Wireless Product Marketing, focusing on BaseStation products in the Communications Business Unit, for Xilinx. He has more than 20 years of experience in the semiconductor industry in roles ranging from Design, Application Engineering, Sales and Marketing. Prior to transferring to marketing from Technical Sales in 2005, he was Senior Staff FAE covering the Wireless market sector in the UK for 7 years.

During his career, David has been published in a number of trade journals and spoken at a number of conferences and industry events such as LTE World Summit, Wireless China, Next Generation Networks and IWPC conferences. David began his career in design at Rutherford Appleton Laboratory, UK. David holds a B.Eng in Electronics from De Montfort University, Leicester, UK.

Xilinx is the world’s leading provider of All Programmable FPGAs, SoCs and 3D ICs. These industry-leading devices are coupled with a next-generation design environment and IP to serve a broad range of customer needs, from programmable logic to programmable systems integration.

Most popular articles in Antennas & propagation

  • Beam forming for 5G communication systems
  • Indoor Wireless Coverage: the Challenges
  • Antenna Technology in the LTE Era
  • Practical Applications for Distributed Antenna Systems
  • MIMO Signal Processing Considerations
  • Share this page

    Want more like this? Register for our newsletter

    What makes e-paper the best display technology for Makers? Scott Soong | Pervasive Displays
    What makes e-paper the best display technology for Makers?
    Scott Sonng or Pervasive Displays discusses how e-paper technology is contributing to the world of makers rather than just major companies enabling makers to utilise its advantages in projects based around Raspberry Pi and other single board computers.
    Low Loss Dynamic Compression of CPRI Baseband Data
    Read this paper from Altera that describes a method of using Mu-Law compression for Gaussian-like waveforms providing an efficient methodology.

    More whitepapers

    Radio-Electronics.com is operated and owned by Adrio Communications Ltd and edited by Ian Poole. All information is © Adrio Communications Ltd and may not be copied except for individual personal use. This includes copying material in whatever form into website pages. While every effort is made to ensure the accuracy of the information on Radio-Electronics.com, no liability is accepted for any consequences of using it. This site uses cookies. By using this site, these terms including the use of cookies are accepted. More explanation can be found in our Privacy Policy