FPGAs offer cost-effective, flexible solutions for remote radio heads
The explosion of smartphone and tablets is putting exponential strain on wireless networks. Whereas smartphones took about eight years to reach a 10 percent market adoption point, usage jumped to 40 percent in less than three years following the introduction of the iPhone in 2007, according to MIT’s quarterly Technology Review. That rapid growth continues. This user demand has pushed network operators and the OEMs supplying their equipment to cover a broad spectrum of wireless technologies – including GSM, CDMA2000, TD-SCDMA, UMTS and LTE – all while boosting capacity by using new frequencies, higher bandwidth and greater numbers of cell sites to meet the increasing demands.
Competitive pressures and rapid demand are forcing a need to get new products into market at ever faster rates. The newest generation of field programmable gate arrays (FPGAs) – manufactured at the proven 28nm process technology node – offer tighter integration, a reduced BOM costs and increased operational efficiency. By using off-the-shelf intellectual property (IP) and Xilinx 7 series FPGAs and Zynq-7000 All Programmable SoC devices, OEMs can meet shifting market demands while avoiding the huge upfront investment of $20 million or more required to spin a new, fixed architecture device. These FPGA and SoC devices are tightly integrated to deliver more system functions with a focus on programmability, the highest bandwidth and parallel processing available to leverage programmable logic, fewer components, lower power consumption and accelerated design productivity.
Growing use of remote radio heads
While simultaneously densifying packet throughput for both voice and data in packed urban corridors, operators have been scrambling to extend radio coverage to rural areas, inside buildings and through tunnels, by connecting BTS/NodeB/eNodeB via fiber optical cable using common public radio interface protocols. This fiber can be run up to 40km from the base station, simplify multiple-input, multiple-output (MIMO) operation and increase the base stations efficiency while helping to cover gaps in network coverage. Adding antennas to support MIMO techniques is costly but necessary for operators. To reduce the impact on the operational costs associated with the radio mast, equipment manufacturers are looking at methods to reduce the equipment’s footprint – that is, its volume and weight – while also achieving lower cost and power. They are continually innovating in the radio transmission domain, from antennas and diplexers/triplexers to the radio itself, in an effort to reduce the mast footprint. Adding remote radio heads – which rely on local or cloud-based baseband processing – is one approach that allows network expansion without adding additional base station servers.
FPGAs not just for low volume
In the past, FPGAs have not been considered the optimal option due to a perception that each device was too expensive to meet mass volume production requirements. When the reduced BOM of an integrated FPGA or SoC device and faster time-to-market is factored in, however, FPGAs offer a very competitive solution. OEMs can reduce cost even further by migrating their FPGA designs to non-reprogrammable, lower cost solutions that use identical packaging.
A recent side-by-side comparison by Semico Research compared the competitive advantages of getting to market first with a new solution based on an FPGA implementation versus the 18-month minimum required to design and deploy a new, fixed architecture ASIC. A subsequent comparison was made when migrating the FPGA design to a lower-cost non-reprogrammable solution with the same packaging and footprint. Semico found time-to-market to be a critical success factor because companies that fail to adjust quickly, run the risk of losing market share to competitors and may possibly never make back their initial investment in silicon design costs.
Transition from a flexible proof-of-concept device to a cost-effective production solution
In the past, OEMs have had to confront the valley between developing and proving a new design with flexible FPGAs to maximized production. Where FPGAs allowed designers to develop and deploy working products for customers, growing market share and a desire to reduce device costs often drove a shift to fixed architecture devices. This involved large development costs, time and requalification and re-verification.
For OEMs designing systems based on Kintex-7 and Virtex-7 FPGAs built on the 28nm process node, Xilinx offers EasyPath-7 FPGAs that allow a transition from early development and deployment to production volume devices in as little as six weeks. These devices offer a 35 percent cost reduction without the minimum order quantities (MOQs) required of ASIC spins and production.
Time is money: TTM comparison of remote radio head board applications
Semico recently published a case study where two OEMs deployed new radio head equipment to address market demand. The results were quite dramatic.
Figure 1. Early and late market entry cumulative board
profits: Remote radio head application
Table 1. Remote Radio Head Board Application
Source: Semico Research Corp. and Advance Tech Marketing. Notes: FPGA Unit Price: 1 FPGA per board. ASP $100 (1 x $100= $100); FPGA Design Cost: $200K / person year, 5PY = $1M (6-7 engineers for 9 months); ASIC Design Cost: Labor + NRE (mask, packaging, prototype cost and testing).
As Figure 1 and Table 1 illustrate, the combination of an FPGA design with a switchover to an EasyPath solution at a predetermined point in the production ramp produces a significant increase in the profits that Company #1 has generated compared to Company #2 with their ASIC approach. Since the ASIC design cost (in this case, approximately $20 million) is appreciably more than the combined FPGA and EasyPath solution (estimated at $1 million), it takes Company #2 longer to recoup their design investment. In addition, since the ASIC solution is new and just ramping into production, the yields at the beginning of the silicon production ramp can fluctuate or be unstable. This can put additional pressure on Company #2’s final market share for this application since they may not have the units available to meet market demand as and when it occurs.
The combined FPGA and EasyPath solution helped Company #1 enter the market early. This head start on the competition resulted in Company #1 generating $286 million more in board profits and in shipping 49 percent of the total available units in this market. In addition, the use of this solution helped Company #1 stabilize their profits at a time in the production ramp when the market demand was peaking. Without this cost reduction solution, Company #1 might have missed their profitability targets altogether and suffered substantially lower profits.
Summary
The explosion in demand for ubiquitous high-speed data is driving continued innovation in tower-mounted antenna and digital radio solutions. All of these solutions have one thing in common – the need to be smaller, lighter, lower in cost and in power, but at the same time, highly integrated and with high levels of flexibility to cater to the differing demands of the networks.
For remote radio head installations and several other network challenges, Xilinx’s 28nm FPGA devices offer the unparalleled ability to create products with maximum flexibility, maximum integration and the lowest overall cost, power and weight.
About the author
Sumit Shah is Senior Product Line Manager, High Volume and EasyPath FPGAs, at Xilinx. In this role, Sumit has responsibility for overall product management and go-to-market strategies for Xilinx’s high volume FPGA product lines as well as EasyPath FPGAs, which enable customers to cost-reduce their designs in under six weeks, in a fast, simple, and risk-free way.
Sumit joined Xilinx in 2007 and has over 12 years of marketing and engineering experience in the programmable logic industry. Prior to joining Xilinx, he worked as an engineering system manager in aerospace and defense industries with Raytheon and General Dynamics, where he designed next-generation electronics systems based on programmable logic. Sumit holds a M.A. in Business Economics and a B.S. in Electrical & Computer Engineering from University of California, Santa Barbara.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
