With multiple radio support now required in the radio network, programmable digital radio may provide a method of reducing spending and increasing efficiency for suppliers, vendors and operators, says David Hawke, Radio Product Manager, Xilinx Inc.
With the increasing number of emerging wireless standards and the multitude of possible transmission frequencies across the globe, infrastructure Original Equipment Manufacturers (OEMs) are faced with increasing development and integration costs to satisfy multiple markets. These markets differ substantially depending on the needs and spending power of the consumers.
Today's consumers now expect high speed connectivity in the home, and the natural evolution of this is to make this facility mobile. Making data low cost and enabling wireless Network Operators price plan's appeal to a wide range of clientele is of paramount importance to consumer attractiveness and retention.
In order to provide lower cost data spectral efficiency must be increased, as has been demonstrated with the newer air interface standards such as WiMAX, 3GPP-LTE or UMB. This increase doesn't come without a price, as the newer standards place an even greater linearity requirement on the radio equipment resulting in potentially lower efficiency. This directly affects Capital Expenditures (CapEx) and Operating Expenditures (OpEx). Without help from advanced algorithms and Moore's Law, radio transmission efficiencies would decrease. How can this be minimised whilst allowing Network Operators greater freedom of choice of how they use their spectrum?
Multi-Modal Radio can offer benefits across all levels of the supply. Having a flexible and programmable radio allows Network Operators and OEMs the ability to provide support for new and evolving standards with minimal risk. It also has the ability to provide Operators the ability to plan their coverage depending on the likelihood of the demand from their customer base. Clearly, there is likely to be most return on mobile high speed data close to business centres, transport hubs and urban areas. Other areas may benefit from lower cost voice coverage to compete directly with fixed line to the home.
The concept of Multi-Modal Radio is not new. Military applications have long been pursuing Software Defined Radio (SDR) as a truly reconfigurable platform that can support many different waveforms (or air interface standards).
Advances in Design
Unfortunately, analogue circuits don't tend to follow Moore's Law in the same way digital logic does. However, there are certainly advances that are making a large impact on present and future radio design.
The data converter manufacturers are now able to provide support for Multi-Modal Radio through the use of direct conversion architectures. This architecture allows for a much lower Bill of Materials (BOM) cost, since it removes the need for radio Intermediate Frequencies (IF) found in heterodyne designs. It consumes less board area allowing for more compact units. This architecture allows for a single Local Oscillator (LO) providing the transmission frequency required by the Operator. Multiple switchable LO's allow the same unit to be used over a number of frequencies and wireless standards, resulting in lower integration costs and decreased risk to Operators.
Digital techniques
Digital logic advances have progressed significantly since the first CMOS devices. Programmable logic, in particular Field Programmable Gate Arrays (FPGAs) have evolved from being simple glue to full System on Chip (SoC), but remain programmable.
Integrated SERialiser/DESerialiser (SERDES) allow OBSAI and CPRI radio communication protocols to be implemented in FPGAs. Additionally, a significant increase in the amount of Digital Signal Processing (DSP) blocks allows for efficient implementation of signal processing requirements demanded by radio designs. Such is the integration now possible; the entire digital processing requirements of a radio can be implemented in a single FPGA for lower cost and lower power.
Power Amplifiers (PA's) are well known to be the largest proportion of the cost of running the BaseStation, due in part to the requirement to be highly linear with the later signalling standards. Amplifier design is also progressing, but coupling this with advanced digital algorithms provides significant efficiency improvement with low equipment costs. Two well known methods of improving PA efficiency are Crest Factor Reduction (CFR) and Digital Pre-Distortion (DPD).
Due to the advancement of FPGAs, CFR and DPD algorithms can now be realised cost effectively. Xilinx has seen efficiency improved from 6% to 34% on certain amplifier types with UMTS using it's CFR and DPD algorithms. This significantly reduces the OpEx costs of running network equipment, and contributes to a reduction in CO2 emissions due to energy savings.
Techniques that really make programmable digital radio a reality are already available. Signal processing techniques exist such that any mixture of air interface waveform can be deployed using an FPGA. The inherently programmable nature of FPGAs allows radios to be future proofed, allowing optimisation or functional changes to be made long after network deployment.
Mixing air interfaces
These digital advances, combined with software programmability, means that any signal processing chain can be deployed for a given air interface. If this concept is taken further, then it is possible to mix air interface types. For example, it could be desirable to have a radio configured such that there is a single carrier of UMTS in the proximity of a dense business centre with a single adjacent 10MHz WiMAX carrier in the same transmitted spectrum. This would suit both the need for voice connectivity and low cost data. In an urban area, it may be more practical to have a higher density of UMTS carriers deployed in that same spectrum where the return on investment for the voice traffic may be higher.
One key advantage is that the same radio platform is used for all frequency variants (subject to RF component fit options), and all air interface standards. This in itself provides a streamlined product line for OEMs that allows for reduced integration, verification and qualification testing, resulting in far lower costs than having multiple separate platforms for UMTS at 2.1GHz, WiMAX at 2.5GHz or WiMAX at 3.5GHz for example. All of the above could be supported from the same base PCB and digital circuitry, allowing significant Bill of Materials (BOM) reduction over having multiple platforms with differing base architectures.
Increasing reliability
Maximising a system's reliability is key in reducing OpEx and increasing profitability. Radio electronics reliability is affected by a number of elements. One of those elements is thermal stress.
Thermally, radios are challenging. RF electronics and power amplifiers tend to run hot, and it's this thermal design which can adversely affect a unit's reliability. Reducing component thermal stress on both the analogue domain (RF components and power amplifiers) and digital domain (digital radio processing) results in far superior equipment reliability.
A typical radio subsystem has many hundreds of components, where many of these are in the analogue domain. In newer analogue architectures there is a significant reduction in component count. This reduction can often lead to providing a statistically higher reliability figure. Additionally, integration in the digital domain to a single device from many Application Specific Standard Products (ASSPs) results in lower power consumption. Reducing this consumption allows easier heat dissipation and a lower FPGA junction temperature.
The reliability of Xilinx' 90nm FPGAs are reported as 2 FIT's or Failures In Time. This means that statistically there would be only 2 failures per 1 billion device hours.
Consumer benefits
If Operators and OEMs take advantage of technology advances, it would bring new meaning to network optimisation. Networks could be optimised such that in certain areas, broadband data could be made available on an ‘all you can eat' basis for the business end of the spectrum, whilst also being able to cater to the requirements to be competitive at the voice only end. The network could really be optimised on a cell by cell basis.
Providing services that customers are willing to pay for, by having flexible, efficient and cost optimal radio hardware allows networks to be continually tuned and adapted to the ever-changing market needs. This strategy allows greater profitability and opportunity to increase ARPU depending on the various consumer personas targeted by the operator, whilst mitigating risks associated by deploying completely new networks.